🤖Prompt Engineering: Challenges (Just the Quotes)

"Another problem that can be confusing is that LLMs seldom put out the same thing twice. [...] Traditional databases are straightforward - you ask for something specific, and you get back exactly what was stored. Search engines work similarly, finding existing information. LLMs work differently. They analyze massive amounts of text data to understand statistical patterns in language. The model processes information through multiple layers, each capturing different aspects - from simple word patterns to complex relationships between ideas." (Jeremy C Morgan, "Coding with AI: Examples in Python", 2025)

"Chain-of-thought prompting is a method that forces LLMs to reason through a series of steps, resulting in more structured, transparent, and precise outputs. The goal is to break down complex tasks into smaller, interconnected subtasks, allowing the LLM to address each subtask in a stepby-step manner. This not only helps the model to 'focus' on specific aspects of the problem, but also encourages it to generate intermediate outputs, making it easier to identify and debug potential issues along the way. Another significant advantage of chain-of-thought prompting is the improved interpretability and transparency of the LLM-generated response. By offering insights into the model’s reasoning process, we, as users, can better understand and qualify how the final output was derived, which promotes trust in the model’s decision-making abilities." (Sinan Ozdemir, "Quick Start Guide to Large Language Models: Strategies and Best Practices for Using ChatGPT and Other LLMs", 2024) 

"AI isn’t just going to be about our digital world. It’s also about our physical world; and applied properly, imagine what AI can do for the pace of discovery and innovation. It’s not just makeup; imagine what it can do for new materials discovery for medicine, energy, climate, and all the other pressing challenges we face as a species - these are the same challenges of makeup, just described with a different 'language'. And quantum computing evolves, we’re bound to see a synergy of these innovations that we can use to tackle these problem domains and more." (Rob Thomas et al, "AI Value Creators: Beyond the Generative AI User Mindset", 2025)

"LLMs can inadvertently produce toxic content or biased language, leak private information, or be vulnerable to jailbreak prompts. These risks carry serious legal and reputational consequences. To mitigate them, evaluation tools must integrate automated filters and classifiers that flag problematic outputs in real time, as we discussed earlier in the chapter. Metrics such as safety scores, toxicity indices, and bias measurements should be collected alongside model metadata for auditing purposes." (Abi Aryan, "LLMOps: Managing Large Language Models in Production", 2025)

"LLM developers can train the model simply to perform well on the benchmarks, like a student memorizing the answers to an upcoming exam. This is a very serious problem in practice. It’s not uncommon to see an LLM perform well in general benchmarks, only to perform below the level of GPT-3.5 (a now-obsolete but inexpensive model) in a practical application, like describing a scene. When this happens, there’s usually little reason to use the model that has the higher general scores - your users should have the final word. Another problem is that LLMs are highly sensitive to the compatibility of the data used in training and prompts used in evaluation. A seemingly minor change in the prompt can lead to drastically different outputs. This makes it difficult to design prompts that consistently elicit the desired response and assess the LLM’s true capabilities." (Abi Aryan, "LLMOps: Managing Large Language Models in Production", 2025)

"The art of mega-prompts spanning multiple written pages and looking like essays has become commonplace for complex tasks when building applications to get things `just right'. Unfortunately, they bring with them lots of issues: errors, portability, complexity, and more. The GenAI world didn’t plan for mega-prompts. They have simply evolved into what they’ve become today because practitioners kept wanting to do more and more complex things, and their only way to express those intents was with a prompt. But step back and look at some of these prompts [...] Lurking just below the surface are a bunch of classical computing concepts like data, programming instructions, control flows, memory, and stora - all the components typically associated with classical computing elements." (Rob Thomas et al, "AI Value Creators: Beyond the Generative AI User Mindset", 2025)

"The same difficulties that characterize training deep feedforward networks also apply to RNNs; gradients tend to die out over long distances using traditional activation functions (or explode if the gradients become greater than 1). However, unlike feedforward networks, RNNs aren’t trained with traditional backpropagation, but rather a variant known as Backpropagation through Time (BPTT): the network is unrolled, as before, and backpropagation is used, averaging over errors at each time point (since an 'output', the hidden state, occurs at each step). Also, in the case of RNNs, we run into the problem that the network has a very short memory; it only incorporates information from the most recent unit before the current one and has trouble maintaining long-range context. For applications such as translation, this is clearly a problem, as the interpretation of a word at the end of a sentence may depend on terms near the beginning, not just those directly preceding it." (Joseph Babcock & Raghav Bali, "Generative AI with Python and PyTorch" 2nd. Ed., 2025)

"When there are hidden layers between the input and output, the problem becomes more complex: when do we change the internal weights to compute the activations that feed into the final output? How do we modify them in relation to the input weights? The insight of the backpropagation technique is that we can use the chain rule from calculus to efficiently compute the derivatives of each parameter of a network with respect to a loss function and, combined with a learning rule, this provides a scalable way to train multilayer networks." (Joseph Babcock & Raghav Bali, "Generative AI with Python and PyTorch" 2nd. Ed., 2025)

"There is no law of physics tdictates AI must remain expensive. The cost of training and inference isn’t fixed - it is an engineering challenge to solved. Businesses, both incumbents and upstarts, have the ingenuity to push these costs down and make AI more practical and widespread." (Rob Thomas et al, "AI Value Creators: Beyond the Generative AI User Mindset", 2025)

"While the backpropagation procedure provides a way to update interior weights within the network in a principled way, it has several shortcomings that make deep networks difficult to use in practice. One is the problem of vanishing gradients. [...] As the value of the sigmoid function increases or decreases toward the extremes (0 or 1, representing either 'off' or 'on' ), the values of the gradient vanish to near zero. This means that the updates to and , which are products of these gradients from hidden activation functions , shrink toward zero, making the weights change little between iterations and making the parameters of the hidden layer neurons change very slowly during backpropagation. Clearly, one problem here is that the sigmoid function saturates; thus, choosing another nonlinearity might circumvent this problem." (Joseph Babcock & Raghav Bali, "Generative AI with Python and PyTorch" 2nd. Ed., 2025)

🖍️Andreas C Müller - Collected Quotes

"A major challenge in unsupervised learning is evaluating whether the algorithm learned something useful. Unsupervised learning algorithms are usually applied to data that does not contain any label information, so we don’t know what the right output should be. Therefore, it is very hard to say whether a model 'did well'. [...] As a consequence, unsupervised algorithms are used often in an exploratory setting, when a data scientist wants to understand the data better, rather than as part of a larger automatic system. Another common application for unsupervised algorithms is as a preprocessing step for supervised algorithms. Learning a new representation of the data can sometimes improve the accuracy of supervised algorithms, or can lead to reduced memory and time consumption." (Andreas C Müller & Sarah Guido, "Introduction to Machine Learning with Python: A Guide for Data Scientists", 2017)

"An important property of neural networks is that their weights are set randomly before learning is started, and this random initialization affects the model that is learned. That means that even when using exactly the same parameters, we can obtain very different models when using different random seeds. If the networks are large, and their complexity is chosen properly, this should not affect accuracy too much, but it is worth keeping in mind (particularly for smaller networks)." (Andreas C Müller & Sarah Guido, "Introduction to Machine Learning with Python: A Guide for Data Scientists", 2017)

"[...]  adding nonlinear features to the representation of our data can make linear models much more powerful. However, often we don’t know which features to add, and adding many features (like all possible interactions in a 100-dimensional feature space) might make computation very expensive. Luckily, there is a clever mathematical trick that allows us to learn a classifier in a higher-dimensional space without actually computing the new, possibly very large representation. This is known as the kernel trick, and it works by directly computing the distance (more precisely, the scalar products) of the data points for the expanded feature representation, without ever actually computing the expansion." (Andreas C Müller & Sarah Guido, "Introduction to Machine Learning with Python: A Guide for Data Scientists", 2017)

"Agglomerative clustering produces what is known as a hierarchical clustering. The clustering proceeds iteratively, and every point makes a journey from being a single point cluster to belonging to some final cluster. Each intermediate step provides a clustering of the data (with a different number of clusters). It is sometimes helpful to look at all possible clusterings jointly. [...] While this visualization provides a very detailed view of the hierarchical clustering, it relies on the two-dimensional nature of the data and therefore cannot be used on datasets that have more than two features. There is, however, another tool to visualize hierarchical clustering, called a dendrogram, that can handle multidimensional datasets." (Andreas C Müller & Sarah Guido, "Introduction to Machine Learning with Python: A Guide for Data Scientists", 2017)

"Before building a machine learning model it is often a good idea to inspect the data, to see if the task is easily solvable without machine learning, or if the desired information might not be contained in the data. Additionally, inspecting your data is a good way to find abnormalities and peculiarities. Maybe some of your irises were measured using inches and not centimeters, for example. In the real world, inconsistencies in the data and unexpected measurements are very common." (Andreas C Müller & Sarah Guido, "Introduction to Machine Learning with Python: A Guide for Data Scientists", 2017)

"Decision trees have two advantages over many of the algorithms [...]: the resulting model can easily be visualized and understood by nonexperts (at least for smaller trees), and the algorithms are completely invariant to scaling of the data. As each feature is processed separately, and the possible splits of the data don’t depend on scaling, no preprocessing like normalization or standardization of features is needed for decision tree algorithms. In particular, decision trees work well when you have features that are on completely different scales, or a mix of binary and continuous features. The main downside of decision trees is that even with the use of pre-pruning, they tend to overfit and provide poor generalization performance. Therefore, in most applications, the ensemble methods we discuss next are usually used in place of a single decision tree." (Andreas C Müller & Sarah Guido, "Introduction to Machine Learning with Python: A Guide for Data Scientists", 2017)

"Essentially, random forests share all of the benefits of decision trees, while making up for some of their deficiencies. One reason to still use decision trees is if you need a compact representation of the decision-making process. It is basically impossible to interpret tens or hundreds of trees in detail, and trees in random forests tend to be deeper than decision trees (because of the use of feature subsets). Therefore, if you need to summarize the prediction making in a visual way to nonexperts, a single decision tree might be a better choice. While building random forests on large datasets might be somewhat time consuming, it can be parallelized across multiple CPU cores within a computer easily." (Andreas C Müller & Sarah Guido, "Introduction to Machine Learning with Python: A Guide for Data Scientists", 2017)

"For regression tasks, the goal is to predict a continuous number, or a floating-point number in programming terms (or real number in mathematical terms). Predicting a person’s annual income from their education, their age, and where they live is an example of a regression task. When predicting income, the predicted value is an amount, and can be any number in a given range. [...] An easy way to distinguish between classification and regression tasks is to ask whether there is some kind of continuity in the output. If there is continuity between possible outcomes, then the problem is a regression problem." (Andreas C Müller & Sarah Guido, "Introduction to Machine Learning with Python: A Guide for Data Scientists", 2017)

"Gradient boosted decision trees are among the most powerful and widely used models for supervised learning. Their main drawback is that they require careful tuning of the parameters and may take a long time to train. Similarly to other tree-based models, the algorithm works well without scaling and on a mixture of binary and continuous features. As with other tree-based models, it also often does not work well on high-dimensional sparse data." (Andreas C Müller & Sarah Guido, "Introduction to Machine Learning with Python: A Guide for Data Scientists", 2017)

"How well the uncertainty actually reflects uncertainty in the data depends on the model and the parameters. A model that is more overfitted tends to make more certain predictions, even if they might be wrong. A model with less complexity usually has more uncertainty in its predictions. A model is called calibrated if the reported uncertainty actually matches how correct it is - in a calibrated model, a prediction made with 70% certainty would be correct 70% of the time." (Andreas C Müller & Sarah Guido, "Introduction to Machine Learning with Python: A Guide for Data Scientists", 2017)

"In binary classification we often speak of one class being the positive class and the other class being the negative class. Here, positive doesn’t represent having benefit or value, but rather what the object of the study is. So, when looking for spam, “positive” could mean the spam class. Which of the two classes is called positive is often a subjective matter, and specific to the domain." (Andreas C Müller & Sarah Guido, "Introduction to Machine Learning with Python: A Guide for Data Scientists", 2017)

"It’s important to note that model complexity is intimately tied to the variation of inputs contained in your training dataset: the larger variety of data points your data‐ set contains, the more complex a model you can use without overfitting. Usually, collecting more data points will yield more variety, so larger datasets allow building more complex models. However, simply duplicating the same data points or collecting very similar data will not help." (Andreas C Müller & Sarah Guido, "Introduction to Machine Learning with Python: A Guide for Data Scientists", 2017)

"Kernelized support vector machines are powerful models and perform well on a variety of datasets. SVMs allow for complex decision boundaries, even if the data has only a few features. They work well on low-dimensional and high-dimensional data (i.e., few and many features), but don’t scale very well with the number of samples. Running an SVM on data with up to 10,000 samples might work well, but working with datasets of size 100,000 or more can become challenging in terms of runtime and memory usage. Another downside of SVMs is that they require careful preprocessing of the data and tuning of the parameters. This is why, these days, most people instead use tree-based models such as random forests or gradient boosting (which require little or no pre‐ processing) in many applications. Furthermore, SVM models are hard to inspect; it can be difficult to understand why a particular prediction was made, and it might be tricky to explain the model to a nonexpert." (Andreas C Müller & Sarah Guido, "Introduction to Machine Learning with Python: A Guide for Data Scientists", 2017) 

"Learning a decision tree means learning the sequence of if/else questions that gets us to the true answer most quickly. In the machine learning setting, these questions are called tests (not to be confused with the test set, which is the data we use to test to see how generalizable our model is). Usually data does not come in the form of binary yes/no features as in the animal example, but is instead represented as continuous features [...]. The tests that are used on continuous data are of the form 'Is feature i larger than value a?'" (Andreas C Müller & Sarah Guido, "Introduction to Machine Learning with Python: A Guide for Data Scientists", 2017)

"Linear regression, or ordinary least squares (OLS), is the simplest and most classic linear method for regression. Linear regression finds the parameters w and b that minimize the mean squared error between predictions and the true regression targets, y, on the training set. The mean squared error is the sum of the squared differences between the predictions and the true values. Linear regression has no parameters, which is a benefit, but it also has no way to control model complexity." (Andreas C Müller & Sarah Guido, "Introduction to Machine Learning with Python: A Guide for Data Scientists", 2017)

"Manifold learning algorithms are mainly aimed at visualization, and so are rarely used to generate more than two new features. Some of them, including t-SNE, com‐ pute a new representation of the training data, but don’t allow transformations of new data. This means these algorithms cannot be applied to a test set: rather, they can only transform the data they were trained for. Manifold learning can be useful for exploratory data analysis, but is rarely used if the final goal is supervised learning. The idea behind t-SNE is to find a two-dimensional representation of the data that preserves the distances between points as best as possible. t-SNE starts with a random twodimensional representation for each data point, and then tries to make points that are close in the original feature space closer, and points that are far apart in the original feature space farther apart. t-SNE puts more emphasis on points that are close by, rather than preserving distances between far-apart points. In other words, it tries to preserve the information indicating which points are neighbors to each other." (Andreas C Müller & Sarah Guido, "Introduction to Machine Learning with Python: A Guide for Data Scientists", 2017)

"Neural networks - particularly the large and powerful ones - often take a long time to train. They also require careful preprocessing of the data, as we saw here. Similarly to SVMs, they work best with 'homogeneous' data, where all the features have similar meanings. For data that has very different kinds of features, tree-based models might work better. Tuning neural network parameters is also an art unto itself. In our experiments, we barely scratched the surface of possible ways to adjust neural network models and how to train them."  (Andreas C Müller & Sarah Guido, "Introduction to Machine Learning with Python: A Guide for Data Scientists", 2017)

"Ridge regression is also a linear model for regression, so the formula it uses to make predictions is the same one used for ordinary least squares. In ridge regression, though, the coefficients (w) are chosen not only so that they predict well on the training data, but also to fit an additional constraint. We also want the magnitude of coef‐ficients to be as small as possible; in other words, all entries of w should be close to zero. Intuitively, this means each feature should have as little effect on the outcome as possible (which translates to having a small slope), while still predicting well. This constraint is an example of what is called regularization. Regularization means explicitly restricting a model to avoid overfitting." (Andreas C Müller & Sarah Guido, "Introduction to Machine Learning with Python: A Guide for Data Scientists", 2017)

"The gradient boosted regression tree is another ensemble method that combines multiple decision trees to create a more powerful model. Despite the 'regression' in the name, these models can be used for regression and classification. In contrast to the random forest approach, gradient boosting works by building trees in a serial manner, where each tree tries to correct the mistakes of the previous one. By default, there is no randomization in gradient boosted regression trees; instead, strong pre-pruning is used. Gradient boosted trees often use very shallow trees, of depth one to five, which makes the model smaller in terms of memory and makes predictions faster. The main idea behind gradient boosting is to combine many simple models (in this context known as weak learners), like shallow trees. Each tree can only provide good predictions on part of the data, and so more and more trees are added to iteratively improve performance." (Andreas C Müller & Sarah Guido, "Introduction to Machine Learning with Python: A Guide for Data Scientists", 2017)

"Typically only a subset of the training points matter for defining the decision boundary: the ones that lie on the border between the classes. These are called support vectors and give the support vec‐ tor machine its name. To make a prediction for a new point, the distance to each of the support vectors is measured. A classification decision is made based on the distances to the support vector, and the importance of the support vectors that was learned during training.". (Andreas C Müller & Sarah Guido, "Introduction to Machine Learning with Python: A Guide for Data Scientists", 2017)

"Unsupervised transformations of a dataset are algorithms that create a new representation of the data which might be easier for humans or other machine learning algorithms to understand compared to the original representation of the data. A common application of unsupervised transformations is dimensionality reduction, which takes a high-dimensional representation of the data, consisting of many features, and finds a new way to represent this data that summarizes the essential characteristics with fewer features. A common application for dimensionality reduction is reduction to two dimensions for visualization purposes." (Andreas C Müller & Sarah Guido, "Introduction to Machine Learning with Python: A Guide for Data Scientists", 2017)

🪙Business Intelligence: Data Silos (Just the Quotes)

"Silos are everywhere. […] Silos are common because they are simple, reliable, and unambiguous." (Scott Rosenberg, "Dreaming in Code", 2007)

"Data mart: A subset of a data warehouse that’s usually oriented to a business group or process rather than enterprise-wide views. They have value as part of the overall enterprise data architecture, but can cause problems when they sprout uncontrolled as data silos with their own data definitions, creating data shadow systems." (Rick Sherman, "Business Intelligence Guidebook: From Data Integration to Analytics, 2015)

"Data marts promised to be quicker and cheaper to build, and provided many more benefits - including the benefit of actually being able to finish building them! The data mart was primarily a backlash to the big, cumbersome CDW projects, with the key difference being that its scope was limited to a single business group rather than the entire enterprise. Of course, that shortcut did speed things up, but at the expense of obtaining agreement on consistent data definitions, thereby guaranteeing data silos." (Rick Sherman, "Business Intelligence Guidebook: From Data Integration to Analytics, 2015)

"A data silo is an isolated source of data that is only accessible to a single line of business (LOB) or department. It leads to inefficiencies, wasted resources, and obstacles in the form of incomplete data profiles and the inability to construct deep insights. [...] On the other hand, a data swamp is a large body of data that is ungoverned and unreliable. It is hard to find data and even harder to use it, which is why it's often used out of context. This is the opposite of data silos in the sense that the data is there and has been brought together, but because it has been done without adequate process and policy, it is as good as not being there. That would be a wasted investment." (Anindita Mahapatra, "Simplifying Data Engineering and Analytics with Delta", 2022)

"Since data engineering is such a crucial field, you may be wondering who the main players are and what skill sets they possess. Building a data product involves several folks, all of whom need to come together with seamless handoffs to ensure a successful end product or service is created. It would be a mistake to create silos and increase both the number and complexity of integration points as each additional integration is a potential failure point." (Anindita Mahapatra, "Simplifying Data Engineering and Analytics with Delta", 2022)

"Data silos often start to develop as the gap between data engineering activities and data science activities begins to grow. Data scientists frequently spend the majority of their time creating separate ETL and data pipelines that clean and transform data and prepare it into features for their models. These silos usually develop because the tools and technologies used for data engineering don’t support the same activities for data scientists." (Bennie Haelen & Dan Davis, "Delta Lake: Up and Running - Modern Data Lakehouse Architectures with Delta Lake", 2023)

"Definition of data and AI governance policies, rules, and classifications is critical to break down data silos, allow for a uniform data consumption, and prevent misuse of data. It includes monitoring of compliance and enforcement of data and AI rules and policies on an ongoing basis, as well as ensuring compliance with regulations and laws." (Eberhard Hechler et al, "Data Fabric and Data Mesh Approaches with AI", 2023)

"Where Data Mesh differs from Data Fabric is that it has fixed requirements for the Self-Service platform focused on organizing and managing Data Products by business domain. Another difference is Data Fabric supports managing data as an asset and as a product. A Data Product can be composed of assets that have been governed and managed in a Data Fabric architecture. Data Fabric does not have these fixed requirements, although it inherently supports isolating data and Data Governance enforcement via metadata by business domain. You can think of a Data Mesh Self-Service data platform as supporting separate, independent companies (business domains), although the key criteria are that it does not create data silos and attains data sharing across these companies in a secure, quick, and easy manner. In Data Mesh, Data Products are created and managed by federated business domains and a data platform requires capabilities that enable data and policy federation. This is where a Data Fabric solution can also address Data Mesh’s requirements." (Sonia Mezzetta, "Principles of Data Fabric: Become a data-driven organization by implementing Data Fabric solutions efficiently", 2023)

"Consider data silos. Data silos hinder data accessibility and collaboration, making it difficult to gain a holistic view and leverage the full potential of the available data. They present a real, present, and formidable challenge that almost all data practitioners experience in modern enterprises. Data silos, much like isolated islands in an immense ocean, are repositories of data that are confined within specific departments or systems, disconnected from the broader organizational data landscape. This segregation results in a fragmented data ecosystem, where valuable insights remain untapped, and the collective intelligence of the enterprise is underutilized." (Jean-Georges Perrin & Eric Broda, "Implementing Data Mesh: Principles and Practice to Design, Build, and Implement Data Mesh", 2024)

"Data Mesh advocates for domain-driven ownership of data, enabling individual teams to manage and share their data effectively while aligning with the overall organizational objectives. By embracing this paradigm, enterprises can gradually dismantle the barriers of data silos, paving the way for a more integrated, agile, and data-centric organizational culture." (Jean-Georges Perrin & Eric Broda, "Implementing Data Mesh: Principles and Practice to Design, Build, and Implement Data Mesh", 2024

"Federated computational governance is essential for maintaining consistency and compatibility across the Data Mesh. It ensures that despite the decentralized nature of data ownership, there is a unified framework governing how data is managed, used, and shared. This unified approach is crucial in preventing data silos, ensuring data interoperability, and maintaining the overall integrity of the data ecosystem." (Jean-Georges Perrin & Eric Broda, "Implementing Data Mesh: Principles and Practice to Design, Build, and Implement Data Mesh", 2024)

"Promoting domain-oriented ownership is to combat the common problem of organizational silos. Silos can significantly hinder the free flow of data and expertise, making decision-making and innovation more challenging. We aim to break down these barriers by advocating for domain-oriented ownership and creating a more dynamic and collaborative data management landscape." (Pradeep Menon, "Data Mesh Principles, patterns, architecture, and strategies for data-driven decision making", 2024)

"The ramifications of data silos extend beyond mere inefficiencies; they actively hinder collaboration and innovation within an organization. When data is trapped in silos, it becomes difficult for teams to access the information they need to collaborate effectively. This lack of accessibility and visibility leads to duplicated efforts, inconsistent data practices, and a general sense of organizational disjointedness." (Jean-Georges Perrin & Eric Broda, "Implementing Data Mesh: Principles and Practice to Design, Build, and Implement Data Mesh", 2024)

"[...] organizations are working to unify the data silos that often exist across their data ecosystem. Some are doing this through a centralized physical approach, such as implementing a data lakehouse pattern, which allows them to store and manage all types of data in one place. Many lakehouse implementations such as those from vendors such as Snowflake and Databricks have evolved into what is often referred to as the modern data platform. This is an architectural pattern that combines the lakehouse with tightly integrated tools for data ingestion, transformation, analytics, observability, and governance. This pattern reflects a shift toward cloud-native architectures that are designed to support end-to-end data workflows with scalability and flexibility." (Fern Halper, "Data Makes the World Go 'Round", 2026)

📉Graphical Representation: Dendrogram (Just the Quotes

"Because the cluster solutions grow tree-like (starting with the branches and ending with the trunk) results are often displayed in a graphic called the dendrogram. Horizontal lines indicate linking of two samples or clusters, and thus the vertical axis presents the associated height or similarity as a measure of distance. The samples are arranged in such a way that the branches of the tree do not overlap. Linking of two groups at a large height indicates strong dissimilarity (and vice versa). Therefore, a clear cluster structure would be indicated if observations are linked at a very low height, and the distinct clusters are linked at a greater height (long branches of the tree). Cutting the dendrogram at such a greater height permits assigning the samples to the resulting distinct clusters. Visual inspection of a dendrogram is often helpful in obtaining an initial estimate of the number of clusters for partitioning methods." (Clemens Reimann et al, "Statistical Data Analysis Explained: Applied Environmental Statistics with R", 2008)

"One useful thing to do once you have clustered something hierarchically is to view the dendrogram of the partition. A dendrogram produces a tree-like diagram that shows how the clusters separate as you move further down the hierarchy." (Drew Conway & John Myles White, "Machine Learning for Hackers", 2012)

"Hierarchical clustering creates a hierarchy of clusters which can be represented in a treelike diagram, called a dendrogram. In the dendrogram, units in the same cluster are joined by a horizontal line, with the scale on the y-axis of the dendrogram reflecting a measure of the distances of the units within the cluster. The leaves at the bottom of the dendrogram represent the individual units; leaves are combined to form small branches, small branches are combined into larger branches, until one reaches the trunk or root of the tree that represents a single cluster containing all units. Dendrograms are quite useful as they give us a visual representation of the clusters." (Johannes Ledolter, "Data mining and business analytics with R", 2013)

"Different joining/linkage rules change how the final hierarchical clustering is presented. [...] Since the barrier for merging observations and clusters is lowest with the single linkage approach, the clustering dendrogram may contain chains of clusters as well as clusters that are spread out. The barrier to joining clusters is highest with complete linkage; however, it is possible that an observation is closer to observations in other clusters than the cluster to which it has been assigned. The average linkage approach moderates the tendencies of the single or complete linkage approaches." (Glenn J Myatt & Wayne P Johnson. "Making sense of data I: a practical guide to exploratory data analysis and data mining" 2nd. Ed., 2014)

"Horizontal trees have proved highly efficient for archetypal models such as classification trees, flow charts, mind maps, dendrograms, and, notably, in the display of files on several software applications and operating systems. If you are a computer user, there is a strong chance you have interacted with some version of a horizontal tree - perhaps on a daily basis." (Manuel Lima, "The Book of Trees: Visualizing Branches of Knowledge", 2014)

"Determining the optimal number of clusters in a data set is a fundamental issue in partitioning clustering, such as k-means clustering, which requires the user to specify the number of clusters k to be generated. Unfortunately, there is no definitive answer to this question. The optimal number of clusters is somehow subjective and depends on the method used for measuring similarities and the parameters used for partitioning. A simple and popular solution consists of inspecting the dendrogram produced using hierarchical clustering to see if it suggests a particular number of clusters. Unfortunately, this approach is also subjective." (Alboukadel Kassambara, "Practical Guide To  Cluster Analysis in R: Unsupervised Machine Learning", 2016)

"The dendrogram is a multilevel hierarchy where clusters at one level are joined together to form the clusters at the next levels. This makes it possible to decide the level at which to cut the tree for generating suitable groups of a data objects." (Alboukadel Kassambara, "Practical Guide To  Cluster Analysis in R: Unsupervised Machine Learning", 2016)

"Agglomerative clustering produces what is known as a hierarchical clustering. The clustering proceeds iteratively, and every point makes a journey from being a single point cluster to belonging to some final cluster. Each intermediate step provides a clustering of the data (with a different number of clusters). It is sometimes helpful to look at all possible clusterings jointly. [...] While this visualization provides a very detailed view of the hierarchical clustering, it relies on the two-dimensional nature of the data and therefore cannot be used on datasets that have more than two features. There is, however, another tool to visualize hierarchical clustering, called a dendrogram, that can handle multidimensional datasets." (Andreas C Müller & Sarah Guido, "Introduction to Machine Learning with Python: A Guide for Data Scientists", 2017)

"Hierarchical clustering is comprised of a general family of clustering algorithms that construct nested clusters by successive merging or splitting of data. The hierarchy of clusters is represented as a tree. The tree is usually called a dendrogram. The root of the dendrogram is the single cluster that contains all the samples; the leaves are the clusters containing only one sample each. This is a nice tool, since it can be straightforwardly interpreted: it 'explains' how clusters are formed and visualizes clusters at different scales. The tree that results from the technique shows the similarity between the samples. Partitioning is computed by selecting a cut on the tree at a certain level." (Laura Igual & Santi Seguí, "Introduction to Data Science: A Python Approach to Concepts, Techniques and Applications", 2017)

"The dendrogram provides a visual representation of the relatedness of variables within a cluster. The height of the horizontal lines indicates the degree of difference between branches. The longer the line, the greater the difference." (Richard V. McCarthy et al, "Applying Predictive Analytics: Finding Value in Data", 2019)

"A tree-like structure or a tree diagram, one showing taxonomic (classification-related) relationships, is created. It is known as a dendrogram. It has branches pointing toward categories or classes. It is a pattern formed by a series of splits or segments of a given quantity of data over a set of compartments and the flow components. From this pattern, a description can be made of the profile of data allocation over the set of functional or structural compartments. The merger of clusters is terminated after every data point lies in one single cluster at the top of the tree." (Vinod K Khanna, "Introduction to Machine Learning Algorithms:  Basic Principles and Mathematics", 2026)

"Hierarchical clustering is summarised by a dendrogram, which sequentially shows points being joined to form a cluster, with the corresponding distances. Breaking the data into clusters is done by cutting the dendrogram at the long edges. [...] Plotting the dendrogram in the data space can help you understand how the hierarchical clustering has collected the points together into clusters. You can learn if the algorithm has been confused by nuisance patterns in the data, and how different choices of linkage method affect the result." (Dianne Cook & Ursula Laa, "Interactively Exploring High-Dimensional Data and Models in R", 2026)

"Viewing the dendrograms in high dimensions provides insight into how the algorithm has joined points to clusters. For example, single linkage often has edges leading to a single focal point, which might not yield a useful clustering but might help to identify outliers. If the edges point to multiple focal points, with long edges bridging gaps in the data, the result is more likely yielding a useful clustering." (Dianne Cook & Ursula Laa,  "Interactively Exploring High-Dimensional Data and Models in R", 2026)

🔭Data Science: Graphs (Just the Quotes)

"A semantic network or net represents knowledge as a net-like graph. An idea, event, situation or object almost always has a composite structure; this is represented in a semantic network by a corresponding structure of nodes (drawn as circles or boxes) representing conceptual units, and directed links (drawn as arrows between the nodes) representing the relations between the units. […] An abstract (graph-theoretic) network can be diagrammed, defined mathematically, programmed in a computer, or hard-wired electronically. It becomes semantic when you assign a meaning to each node and link. Unlike specialized networks and diagrams, semantic networks aim to represent any kind of knowledge which can be described in natural language. A semantic network system includes not only the explicitly stored net structure but also methods for automatically deriving from that a much larger structure or body of implied knowledge." (Fritz Lehman, "Semantic Networks", Computers & Mathematics with Applications Vol. 23 (2-5), 1992)

"The essential idea of semantic networks is that the graph-theoretic structure of relations and. abstractions can be used for inference as well as understanding. […] A semantic network is a discrete structure as is any linguistic description. Representation of the continuous 'outside world' with such a structure is necessarily incomplete, and requires decisions as to which information is kept and which is lost." (Fritz Lehman, "Semantic Networks",  Computers & Mathematics with Applications Vol. 23 (2-5), 1992)

"A graph enables us to visualize a relation over a set, which makes the characteristics of relations such as transitivity and symmetry easier to understand. […] Notions such as paths and cycles are key to understanding the more complex and powerful concepts of graph theory. There are many degrees of connectedness that apply to a graph; understanding these types of connectedness enables the engineer to understand the basic properties that can be defined for the graph representing some aspect of his or her system. The concepts of adjacency and reachability are the first steps to understanding the ability of an allocated architecture of a system to execute properly." (Dennis M Buede, "The Engineering Design of Systems: Models and methods", 2009)

"Graphs can embed complex semantic representations in a compact form. As such, modeling data as networks of related entities is a powerful mechanism for analytics, both for visual analyses and machine learning. Part of this power comes from performance advantages of using a graph data structure, and the other part comes from an inherent human ability to intuitively interact with small networks." (Benjamin Bengfort et al, "Applied Text Analysis with Python: Enabling Language-Aware Data Products with Machine Learning", 2018)

"In Exploiting semantic knowledge graphs can support interpretability and explainability of nearly all AI model types (including DL models) by discovering and depicting semantic and non-obvious relationships or depicting an ML model in a simplified and more readable, explainable way., a Data Mesh solution organizes data around business domain owners and transforms relevant data assets (data sources) to data products that can be consumed by distributed business users from various business domains or functions. These data products are created, governed, and used in an autonomous, decentralized, and self-service manner. Self-service capabilities, which we have already referenced as a Data Fabric capability, enable business organizations to entertain a data marketplace with shopping-for-data characteristics." (Eberhard Hechler et al, "Data Fabric and Data Mesh Approaches with AI", 2023)

"[...] a graph is a mathematical model that is used for describing relationships between entities. However, each complex network presents intrinsic properties. Such properties can be measured by particular metrics, and each measure may characterize one or several local and global aspects of the graph." (Aldo Marzullo et al, "Graph Machine Learning" 2nd Ed., 2025)

"Although creating simple subgraphs and merging them is a way to generate new graphs of increasing complexity, networks may also be generated by means of probabilistic models   and/or generative models that let a graph grow by itself. Such graphs usually share   interesting properties with real networks and have long been used to create benchmarks and synthetic graphs, especially in times when the amount of data available was not as overwhelming as today." (Aldo Marzullo et al, "Graph Machine Learning" 2nd Ed., 2025)

"As with many other deep learning-based approaches, another major challenge is in interpretability. While knowledge graphs provide a structured and transparent way to store relationships, LLMs operate as a black box, making it difficult to understand how specific outputs are generated. [...] Data alignment is also a key issue, as structured knowledge graphs and unstructured text data must be carefully preprocessed to ensure consistency.  Differences in data formats, ontology mismatches, and information redundancy can create inefficiencies when integrating these two paradigms. Developing robust pipelines that seamlessly connect graph-based insights with LLM-generated text remains an open challenge." (Aldo Marzullo et al, "Graph Machine Learning" 2nd Ed., 2025)

"Graph analytics is generally very effective in clustering users, merchants, and communities to provide an effective implementation of behavior analytics. On the other hand, second-party fraud can be identified with the implementation of monitoring employee behavior, as well as compliance checks. Graph analytics can indeed be useful for these use cases. Similar to the first-party models, employee behavior can also be analyzed using graph machine learning, although the dataset may need to encode a number of other sources of information besides transactional data. From a compliance standpoint, process mining techniques that still rely on a graph representation of the various procedural steps/pathways can be effective in identifying fraudulent behavior or non-compliant processes. Finally, third-party fraud, especially in the form of phishing attacks, can also be addressed using graph machine learning. In this context, understanding the network from which the phishing attack comes as well as the URLs being used (which can also benefit from a graph representation) can be critical for building an effective phishing classifier." (Aldo Marzullo et al, "Graph Machine Learning" 2nd Ed., 2025)

"The concept of temporal graphs is useful in all the real-world problems that can be represented as a graph, where the nodes and edges of the graph may change over time. For example, temporal graphs are extensively applied in modeling social networks. By capturing the evolving relationships between individuals, temporal graphs enable a more accurate representation of social dynamics. This is particularly useful for predicting changes in friendships, community structures, and the information diffusion over time." (Aldo Marzullo et al, "Graph Machine Learning" 2nd Ed., 2025)

🔭Data Science: Gradients (Just the Quotes)

"There are many control parameters to a learning system. The question is to identify, at a sufficiently high level, the ones that can play a key role in sequencing effects. Because learning can be seen as the search for an optimal hypothesis in a given space under an inductive criteria defined over the training set, three means to control learning readily appear. The first one corresponds to a change of the hypothesis space. The second consists in modifying the optimization landscape. This can be done by changing either the training set (for instance, by a forgetting mechanism) or the inductive criteria. Finally, one can also fiddle with the exploration process. For instance, in the case of a gradient search, slowing down the search process can prevent the system from having time to find the local optimum, which, in turn, can introduce sequencing effects." (Antoine Cornuéjol, "The Necessity of Order in Machine Learning: Is Order in Order?", 2007)

"Deep learning is about using a stacked hierarchy of feature detectors. [...] we use pattern detectors and we build them into networks that are arranged in hundreds of layers and then we adjust the links between these layers, usually using some kind of gradient descent." (Joscha Bach,Joscha: Computational Meta-Psychology", 2015)

"Boosting is a non-linear flexible regression technique that helps increase the accuracy of trees by assigning more weights to wrong predictions. The reason for inducing more weight is so the model can emphasize more on these wrongly predicted samples and tune itself to increase accuracy. The gradient boosting method solves the inherent problem in boosting trees" (i.e., low speed and human interpretability). The algorithm supports parallelism by specifying the number of threads." (Danish Haroon,Python Machine Learning Case Studies", 2017)

"In Boosting, the selection of samples is done by giving more and more weight to hard-to-classify observations. Gradient boosting classification produces a prediction model in the form of an ensemble of weak predictive models, usually decision trees. It generalizes the model by optimizing for the arbitrary differentiable loss function. At each stage, regression trees fit on the negative gradient of binomial or multinomial deviance loss function." (Danish Haroon,Python Machine Learning Case Studies", 2017)

"The beauty of quantum machine learning is that we do not need to depend on an algorithm like gradient descent or convex objective function. The objective function can be nonconvex or something else." (Amit Ray,Quantum Computing Algorithms for Artificial Intelligence", 2018)

"The process of defining a custom loss function calls for a clear understanding of the objective of your task and the nature of your data. This requires understanding how your model learns and how its predictions can be compared to the actual targets in a meaningful and helpful way. Additionally, it’s crucial to consider the balance between complexity and interpretability of your loss function. While complex functions might capture the task’s intricacies better, they might also make training more challenging and results harder to interpret. At a lower level, we also have to make sure that a custom loss function is differentiable - that is, it must have a derivative everywhere. This requirement arises because learning in these models is accomplished through gradient descent, which requires computing the derivative of the loss function." (Sinan Ozdemir, "Quick Start Guide to Large Language Models: Strategies and Best Practices for Using ChatGPT and Other LLMs", 2024) 

"Transformers are complex models built like LEGO blocks using multiple smart and specialized components. [...] Briefly, a vanilla transformer model consists of separate stacks of encoders and decoders. Each encoder block includes multi-head self-attention, enabling the model to capture relationships between tokens regardless of their positions. Residual connections help maintain gradient flow, preventing the vanishing gradient problem. Layer normalization ensures training stability, and feed-forward layers introduce non-linearity and learn complex token interactions. Decoder blocks contain the same components but also include an encoder-decoder attention mechanism to incorporate context from the encoder. The model uses embedding layers to convert tokens into a continuous latent space for contextual learning and positional encoding to preserve the order of tokens in the sequence." (Joseph Babcock & Raghav Bali, "Generative AI with Python and PyTorch" 2nd. Ed., 2025)

"While the backpropagation procedure provides a way to update interior weights within the network in a principled way, it has several shortcomings that make deep networks difficult to use in practice. One is the problem of vanishing gradients. [...] As the value of the sigmoid function increases or decreases toward the extremes (0 or 1, representing either 'off' or 'on' ), the values of the gradient vanish to near zero. This means that the updates to and , which are products of these gradients from hidden activation functions , shrink toward zero, making the weights change little between iterations and making the parameters of the hidden layer neurons change very slowly during backpropagation. Clearly, one problem here is that the sigmoid function saturates; thus, choosing another nonlinearity might circumvent this problem." (Joseph Babcock & Raghav Bali, "Generative AI with Python and PyTorch" 2nd. Ed., 2025)

"The same difficulties that characterize training deep feedforward networks also apply to RNNs; gradients tend to die out over long distances using traditional activation functions (or explode if the gradients become greater than 1). However, unlike feedforward networks, RNNs aren’t trained with traditional backpropagation, but rather a variant known as Backpropagation through Time (BPTT): the network is unrolled, as before, and backpropagation is used, averaging over errors at each time point (since an 'output', the hidden state, occurs at each step). Also, in the case of RNNs, we run into the problem that the network has a very short memory; it only incorporates information from the most recent unit before the current one and has trouble maintaining long-range context. For applications such as translation, this is clearly a problem, as the interpretation of a word at the end of a sentence may depend on terms near the beginning, not just those directly preceding it." (Joseph Babcock & Raghav Bali, "Generative AI with Python and PyTorch" 2nd. Ed., 2025)

🪙Business Intelligence: Persuation (Just the Quotes)

"In many presentations it is not a question of saving time to the reader but a question of placing the arguments in such form that results may surely be obtained. For matters affecting public welfare, it is hard to estimate the benefits which may accrue if a little care be used in presenting data so that they will be convincing to the reader." (Willard C Brinton, "Graphic Methods for Presenting Facts", 1919)

"There is a magic in graphs. The profile of a curve reveals in a flash a whole situation - the life history of an epidemic, a panic, or an era of prosperity. The curve informs the mind, awakens the imagination, convinces." (Henry D Hubbard [in William Brinton's "Graphic Presentation", 1939])

"Charts and graphs are a method of organizing information for a unique purpose. The purpose may be to inform, to persuade, to obtain a clear understanding of certain facts, or to focus information and attention on a particular problem. The information contained in charts and graphs must, obviously, be relevant to the purpose. For decision-making purposes. information must be focused clearly on the issue or issues requiring attention. The need is not simply for 'information', but for structured information, clearly presented and narrowed to fit a distinctive decision-making context. An advantage of having a 'formula' or 'model' appropriate to a given situation is that the formula indicates what kind of information is needed to obtain a solution or answer to a specific problem." (Cecil H Meyers, "Handbook of Basic Graphs: A modern approach", 1970)

"The more complex the shape of any object. the more difficult it is to perceive it. The nature of thought based on the visual apprehension of objective forms suggests, therefore, the necessity to keep all graphics as simple as possible. Otherwise, their meaning will be lost or ambiguous, and the ability to convey the intended information and to persuade will be inhibited." (Robert Lefferts, "Elements of Graphics: How to prepare charts and graphs for effective reports", 1981)

"To see is to reason. Thus, the use of visual forms of communication has great potential for influencing what a person thinks. Graphic presentation is always much more than a way to present just facts or information. Rather, it is a way to influence thought, and, as such, graphics can be a powerful mode of persuasion." (Robert Lefferts, "Elements of Graphics: How to prepare charts and graphs for effective reports", 1981)

"Always bear in mind that the purposes of any chart are (1) to help gather, organize or visualize the facts; (2) to aid in analyzing them; (3) to help in developing the better method and evaluating it; (4) to assist in convincing management of the improvement’s value." (Ben B Graham, "Detail Process Charting: Speaking the Language of Process", 2004)

"We need [graphic] techniques because figures do not speak for them. selves. Numbers alone seldom make a convincing case or polish their author's image - the twin goals of that other great mind bender, rhetoric. While rhetoric deals in qualitative argument, its quantitative equivalent is graphics. As rhetoric has declined in popularity, so graphics have risen along with our acceptance of quantitative arguments. In graphics, figures finally find their own means of expression." (Nicholas Strange, "Smoke and Mirrors: How to bend facts and figures to your advantage", 2007)

"Oftentimes a statistical graphic provides the evidence for a plausible story, and the evidence, though perhaps only circumstantial, can be quite convincing. […] But such graphical arguments are not always valid. Knowledge of the underlying phenomena and additional facts may be required." (Howard Wainer, "Graphic Discovery: A trout in the milk and other visuals" 2nd, 2008)

"A persuasive visualization primarily serves the relationship between the designer and the reader. It is useful when the designer wishes to change the reader’s mind about something. It represents a very specific point of view, and advocates a change of opinion or action on the part of the reader. In this category of visualization, the data represented is specifically chosen for the purpose of supporting the designer’s point of view, and is presented carefully so as to convince the reader of same." (Noah Iliinsky & Julie Steel, "Designing Data Visualizations", 2011)

"An informative visualization primarily serves the relationship between the reader and the data. It aims for a neutral presentation of the facts in such a way that will educate the reader (though not necessarily persuade him). Informative visualizations are often associated with broad data sets, and seek to distill the content into a manageably consumable form." (Noah Iliinsky & Julie Steel, "Designing Data Visualizations", 2011)

"If you do succeed in persuading them, you’ve only done so on an intellectual basis. That’s not good enough, because people are not inspired to act by reason alone." (Cole N Knaflic, "Storytelling with Data: A Data Visualization Guide for Business Professionals", 2015)

"First, from an ethos perspective, the success of your data story will be shaped by your own credibility and the trustworthiness of your data. Second, because your data story is based on facts and figures, the logos appeal will be integral to your message. Third, as you weave the data into a convincing narrative, the pathos or emotional appeal makes your message more engaging. Fourth, having a visualized insight at the core of your message adds the telos appeal, as it sharpens the focus and purpose of your communication. Fifth, when you share a relevant data story with the right audience at the right time (kairos), your message can be a powerful catalyst for change." (Brent Dykes, "Effective Data Storytelling: How to Drive Change with Data, Narrative and Visuals", 2019)

"The second rule of communication is to know what you want to achieve. Hopefully the aim is to encourage open debate, and informed decision-making. But there seems no harm in repeating yet again that numbers do not speak for themselves; the context, language and graphic design all contribute to the way the communication is received. We have to acknowledge we are telling a story, and it is inevitable that people will make comparisons and judgements, no matter how much we only want to inform and not persuade. All we can do is try to pre-empt inappropriate gut reactions by design or warning." (David Spiegelhalter, "The Art of Statistics: Learning from Data", 2019)

"Data visualization isn’t just about informing, it’s also about persuading." (Steve Wexler, "The Big Picture: How to use data visualization to make better decisions - faster", 2021)

"The lack of focus and commitment to color is a perplexing thing. When used correctly, color has no equal as a visualization tool - in advertising, in branding, in getting the message across to any audience you seek. Data analysts can make numbers dance and sing on command, but they sometimes struggle to create visually stimulating environments that convince the intended audience to tap their feet in time." (Kate Strachnyi, "ColorWise: A Data Storyteller’s Guide to the Intentional Use of Color", 2023)

🤖Prompt Engineering: Training (Just the Quotes)

"[...] building an effective LLM-based application can require more than just plugging in a pre-trained model and retrieving results - what if we want to parse them for a better user experience? We might also want to lean on the learnings of massively large language models to help complete the loop and create a useful end-to-end LLM-based application. This is where prompt engineering comes into the picture." (Sinan Ozdemir, "Quick Start Guide to Large Language Models: Strategies and Best Practices for Using ChatGPT and Other LLMs", 2024)

"Fine-tuning involves training the LLM on a smaller, task-specific dataset to adjust its parameters for the specific task at hand. This allows the LLM to leverage its pre-trained knowledge of the language to improve its accuracy for the specific task. Fine-tuning has been shown to drastically improve performance on domain-specific and task-specific tasks and lets LLMs adapt quickly to a wide variety of NLP applications." (Sinan Ozdemir, "Quick Start Guide to Large Language Models: Strategies and Best Practices for Using ChatGPT and Other LLMs", 2024)

"[...] LLMs are pre-trained on large corpora and sometimes fine-tuned on smaller datasets for specific tasks. Recall that one of the factors behind the Transformer’s effectiveness as a language model is that it is highly parallelizable, allowing for faster training and efficient processing of text. What really sets the Transformer apart from other deep learning architectures is its ability to capture long-range dependencies and relationships between tokens using attention. In other words, attention is a crucial component of Transformer-based LLMs, and it enables them to effectively retain information between training loops and tasks (i.e., transfer learning), while being able to process lengthy swatches of text with ease." (Sinan Ozdemir, "Quick Start Guide to Large Language Models: Strategies and Best Practices for Using ChatGPT and Other LLMs", 2024) 

"The idea behind transfer learning is that the pre-trained model has already learned a lot of information about the language and relationships between words, and this information can be used as a starting point to improve performance on a new task. Transfer learning allows LLMs to be fine-tuned for specific tasks with much smaller amounts of task-specific data than would be required if the model were trained from scratch. This greatly reduces the amount of time and resources needed to train LLMs." (Sinan Ozdemir, "Quick Start Guide to Large Language Models: Strategies and Best Practices for Using ChatGPT and Other LLMs", 2024)

"Transfer learning is a technique used in machine learning to leverage the knowledge gained from one task to improve performance on another related task. Transfer learning for LLMs involves taking an LLM that has been pre-trained on one corpus of text data and then fine-tuning it for a specific 'downstream' task, such as text classification or text generation, by updating themodel’s parameters with task-specific data." (Sinan Ozdemir, "Quick Start Guide to Large Language Models: Strategies and Best Practices for Using ChatGPT and Other LLMs", 2024)

"Data drift manifests in several distinct ways. Input drift typically shows up as an increase in adversarial or malformed queries that deviate from the original training or design expectations. This can stress the system’s robustness and degrade output quality. Retriever drift occurs when the relevance of the documents returned by retrieval components declines, even if the retrieval algorithms and configurations remain unchanged. Similarly, embedding drift arises when the vector representations used to compare semantic similarity become less effective, causing retrieval systems to fail despite stable system parameters." (Abi Aryan, "LLMOps: Managing Large Language Models in Production", 2025)

"Despite their impressive capabilities, LLMs are not without limitations. One of the most significant challenges is the problem of hallucination, where an LLM generates factually incorrect or misleading information that appears plausible. This is particularly problematic in domains requiring high factual accuracy, such as healthcare, finance, and legal applications. To mitigate hallucinations and enhance the reliability of LLM outputs,  Retrieval-Augmented Generation (RAG) has emerged as a powerful technique. RAG works by dynamically retrieving relevant information from an external knowledge source (such as a knowledge graph) at inference time, rather than just relying on pre-trained knowledge. This approach ensures that the model has access to up-to-date and accurate data, grounding answers in verified information rather than generating content purely from its internal representations." (Aldo Marzullo et al, "Graph Machine Learning" 2nd Ed., 2025)

"Generative AI for coding and language tools is based on the LLM concept. A large language model is a type of neural network that processes and generates text in a humanlike way. It does this by being trained on a massive dataset of text, which allows it to learn human language patterns, as described previously. It lets LLMs translate, write, and answer questions with text. LLMs can contain natural language, source code, and  more." (Jeremy C Morgan, "Coding with AI: Examples in Python", 2025)

"In a nutshell, training LLMs involves optimizing a large number of parameters on very large datasets. This process, known as pretraining, typically employs unsupervised learning objectives, such as predicting missing words in a sentence (masked language modeling) or forecasting subsequent words (causal language modeling). As a side effect, the pre-training phase lets the model learn and 'understand' a language, resulting in a remarkable ability to generalize across various tasks, often achieving state-of-the-art performance. LLMs have demonstrated proficiency in a diverse array of applications, reflecting their versatility and depth of language understanding. Key areas include text generation, language translation, question answering, and summarization, among many others." (Aldo Marzullo et al, "Graph Machine Learning" 2nd Ed., 2025)

"LLMs are trained on large volumes of data, which inherently provides them with an immense knowledge base and understanding of different languages. Yet, LLMs at their core are complex text completion engines. Since this knowledge and understanding of language is compressed in a very high-dimensional latent space. LLMs end up using these in a very fluid and intelligible way (which often leads to hallucinations). In order to guide LLMs to focus on specific topics or pieces of information to solve certain tasks, (for instance, question-answering from a given piece of text), it is important to provide contextual information explicitly. While most current generations of LLMs have extremely wide context windows, it is recommended to preprocess context into overlapping smaller chunks for better results, reduced latency, and so on. For similar reasons, it is also recommended to preprocess contextual information in clear and task-specific formats. This aspect of context preprocessing is extremely useful in Retrieval-Gugmented Generation (RAG) scenarios." (Joseph Babcock & Raghav Bali, "Generative AI with Python and PyTorch" 2nd. Ed., 2025)

"The pre-training step is by far the biggest in terms of data and compute requirements for the whole of the LLM’s lifecycle. Yet fine-tuning is quite resource-intensive when we compare it to traditional machine learning and deep learning workflows. Fine-tuning is also a very important step in improving the quality of the models; hence, it makes sense to understand how we can optimize this step without impacting the performance. Efficiencies in this step also enable us to iterate faster, thereby improving adaptability in many fast-moving domains. In this section, we will focus on some interesting efficient method." (Joseph Babcock & Raghav Bali, "Generative AI with Python and PyTorch" 2nd. Ed., 2025)

"Transformers are complex models built like LEGO blocks using multiple smart and specialized components. [...] Briefly, a vanilla transformer model consists of separate stacks of encoders and decoders. Each encoder block includes multi-head self-attention, enabling the model to capture relationships between tokens regardless of their positions. Residual connections help maintain gradient flow, preventing the vanishing gradient problem. Layer normalization ensures training stability, and feed-forward layers introduce non-linearity and learn complex token interactions. Decoder blocks contain the same components but also include an encoder-decoder attention mechanism to incorporate context from the encoder. The model uses embedding layers to convert tokens into a continuous latent space for contextual learning and positional encoding to preserve the order of tokens in the sequence." (Joseph Babcock & Raghav Bali, "Generative AI with Python and PyTorch" 2nd. Ed., 2025)

🖍️Joseph Babcock - Collected Quotes

"An important capability for our LLM app to become smarter is to maintain a working memory of its interactions with us - otherwise, it will approach each prompt with no knowledge of our previous interactions. For example, it won’t remember details like where we live or what our interests are, which would make it more challenging to develop useful LLM assistants that can use personal information about us to provide more engaging, relevant responses. It also makes it practically more challenging to code a personalized application if we have to explicitly pass context for this personalized information with each interaction, rather than maintaining it 'for free' through LangChain’s memory functionality. It can also allow us to make the LLM specialized for different users by maintaining different memories on different 'threads' that we can visualize and retrieve from LangSmith."(Joseph Babcock & Raghav Bali, "Generative AI with Python and PyTorch" 2nd. Ed., 2025)

"At the simplest level, a model, be it machine learning or a more classical method such as linear regression, is a mathematical description of how a target variable changes in response to variation in a predictive variable; that relationship could be a linear slope or any of a number of more complex mathematical transformations." (Joseph Babcock & Raghav Bali, "Generative AI with Python and PyTorch" 2nd. Ed., 2025)

"Data efficiency in LLMs is about maximizing the quality of learning from the available data while minimizing the required dataset size and computational resources. Large datasets are costly to process, and redundant or noisy data can negatively impact model performance. Therefore, data efficiency techniques aim to achieve high model accuracy and generalization with a reduced or optimized dataset. This process includes filtering data for quality, reducing redundancy, and applying sampling techniques to emphasize high-value samples." (Joseph Babcock & Raghav Bali, "Generative AI with Python and PyTorch" 2nd. Ed., 2025)

"Interpretability is an important requirement when it comes to NLP tasks. For computer vision use cases, visual cues are good enough indicators for understanding how a model perceives or generates outputs (quantification is also a problem there, but we can skip it for now). For NLP tasks, since the textual data is first required to be transformed into a vector, it is important to understand what those vectors capture and how they are used by the models." (Joseph Babcock & Raghav Bali, "Generative AI with Python and PyTorch" 2nd. Ed., 2025)

"LLMs are great at generating responses while following instructions but a general empirical observation is a marked improvement in performance when prompts are coupled with a few examples (as opposed to zero-shot scenarios). This is not to say that zero-shot performance is bad but the fact that, in real-life settings, our tasks/requirements are generally a bit more nuanced. For instance, LLMs have an inherent capability to infer sentiment for an input sentence but giving a few examples of how to use that inferred sentiment in responding to customer feedback helps." (Joseph Babcock & Raghav Bali, "Generative AI with Python and PyTorch" 2nd. Ed., 2025)

"LLMs are trained on large volumes of data, which inherently provides them with an immense knowledge base and understanding of different languages. Yet, LLMs at their core are complex text completion engines. Since this knowledge and understanding of language is compressed in a very high-dimensional latent space. LLMs end up using these in a very fluid and intelligible way (which often leads to hallucinations). In order to guide LLMs to focus on specific topics or pieces of information to solve certain tasks, (for instance, question-answering from a given piece of text), it is important to provide contextual information explicitly. While most current generations of LLMs have extremely wide context windows, it is recommended to preprocess context into overlapping smaller chunks for better results, reduced latency, and so on. For similar reasons, it is also recommended to preprocess contextual information in clear and task-specific formats. This aspect of context preprocessing is extremely useful in Retrieval-Gugmented Generation (RAG) scenarios." (Joseph Babcock & Raghav Bali, "Generative AI with Python and PyTorch" 2nd. Ed., 2025)

"[...] simply put, prompt engineering is the practice of designing and refining prompts to guide generative models, particularly LLMs, to produce desired outputs. A prompt is the input to these models, often in plain language, consisting of task instructions (implicit or explicit) with or without examples, enabling users to tap into the model’s vast capabilities." (Joseph Babcock & Raghav Bali, "Generative AI with Python and PyTorch" 2nd. Ed., 2025)

"The pre-training step is by far the biggest in terms of data and compute requirements for the whole of the LLM’s lifecycle. Yet fine-tuning is quite resource-intensive when we compare it to traditional machine learning and deep learning workflows. Fine-tuning is also a very important step in improving the quality of the models; hence, it makes sense to understand how we can optimize this step without impacting the performance. Efficiencies in this step also enable us to iterate faster, thereby improving adaptability in many fast-moving domains. In this section, we will focus on some interesting efficient method." (Joseph Babcock & Raghav Bali, "Generative AI with Python and PyTorch" 2nd. Ed., 2025)

"Transformers are complex models built like LEGO blocks using multiple smart and specialized components. [...] Briefly, a vanilla transformer model consists of separate stacks of encoders and decoders. Each encoder block includes multi-head self-attention, enabling the model to capture relationships between tokens regardless of their positions. Residual connections help maintain gradient flow, preventing the vanishing gradient problem. Layer normalization ensures training stability, and feed-forward layers introduce non-linearity and learn complex token interactions. Decoder blocks contain the same components but also include an encoder-decoder attention mechanism to incorporate context from the encoder. The model uses embedding layers to convert tokens into a continuous latent space for contextual learning and positional encoding to preserve the order of tokens in the sequence." (Joseph Babcock & Raghav Bali, "Generative AI with Python and PyTorch" 2nd. Ed., 2025)

"When there are hidden layers between the input and output, the problem becomes more complex: when do we change the internal weights to compute the activations that feed into the final output? How do we modify them in relation to the input weights? The insight of the backpropagation technique is that we can use the chain rule from calculus to efficiently compute the derivatives of each parameter of a network with respect to a loss function and, combined with a learning rule, this provides a scalable way to train multilayer networks." (Joseph Babcock & Raghav Bali, "Generative AI with Python and PyTorch" 2nd. Ed., 2025)

"While the backpropagation procedure provides a way to update interior weights within the network in a principled way, it has several shortcomings that make deep networks difficult to use in practice. One is the problem of vanishing gradients. [...] As the value of the sigmoid function increases or decreases toward the extremes (0 or 1, representing either 'off' or 'on' ), the values of the gradient vanish to near zero. This means that the updates to and , which are products of these gradients from hidden activation functions , shrink toward zero, making the weights change little between iterations and making the parameters of the hidden layer neurons change very slowly during backpropagation. Clearly, one problem here is that the sigmoid function saturates; thus, choosing another nonlinearity might circumvent this problem." (Joseph Babcock & Raghav Bali, "Generative AI with Python and PyTorch" 2nd. Ed., 2025)

"The same difficulties that characterize training deep feedforward networks also apply to RNNs; gradients tend to die out over long distances using traditional activation functions (or explode if the gradients become greater than 1). However, unlike feedforward networks, RNNs aren’t trained with traditional backpropagation, but rather a variant known as Backpropagation through Time (BPTT): the network is unrolled, as before, and backpropagation is used, averaging over errors at each time point (since an 'output', the hidden state, occurs at each step). Also, in the case of RNNs, we run into the problem that the network has a very short memory; it only incorporates information from the most recent unit before the current one and has trouble maintaining long-range context. For applications such as translation, this is clearly a problem, as the interpretation of a word at the end of a sentence may depend on terms near the beginning, not just those directly preceding it." (Joseph Babcock & Raghav Bali, "Generative AI with Python and PyTorch" 2nd. Ed., 2025)

🤖Prompt Engineering: Performance (Just the Quotes)

"The no free lunch theorem for machine learning states that, averaged over all possible data generating distributions, every classification algorithm has the same error rate when classifying previously unobserved points. In other words, in some sense, no machine learning algorithm is universally any better than any other. The most sophisticated algorithm we can conceive of has the same average performance (over all possible tasks) as merely predicting that every point belongs to the same class. [...] the goal of machine learning research is not to seek a universal learning algorithm or the absolute best learning algorithm. Instead, our goal is to understand what kinds of distributions are relevant to the 'real world' that an AI agent experiences, and what kinds of machine learning algorithms perform well on data drawn from the kinds of data generating distributions we care about." (Ian Goodfellow et al, "Deep Learning", 2015)

"Attention is a mechanism used in deep learning models (not just Transformers) that assigns different weights to different parts of the input, allowing the model to prioritize and emphasize the most important information while performing tasks like translation or summarization. Essentially, attention allows a model to 'focus' on different parts of the input dynamically, leading to improved performance and more accurate results. Before the popularization of attention, most neural networks processed all inputs equally and the models relied on a fixed representation of the input to make predictions. Modern LLMs that rely on attention can dynamically focus on different parts of input sequences, allowing them to weigh the importance of each part in making predictions." (Sinan Ozdemir, "Quick Start Guide to Large Language Models: Strategies and Best Practices for Using ChatGPT and Other LLMs", 2024)

"Different algorithms may perform better on different types of text data and will have different vector sizes. The choice of algorithm can have a significant impact on the quality of the resulting embeddings. Additionally, open-source alternatives may require more customization and finetuning than closed-source products, but they also provide greater flexibility and control over the embedding process." (Sinan Ozdemir, "Quick Start Guide to Large Language Models: Strategies and Best Practices for Using ChatGPT and Other LLMs", 2024)

"Fine-tuning involves training the LLM on a smaller, task-specific dataset to adjust its parameters for the specific task at hand. This allows the LLM to leverage its pre-trained knowledge of the language to improve its accuracy for the specific task. Fine-tuning has been shown to drastically improve performance on domain-specific and task-specific tasks and lets LLMs adapt quickly to a wide variety of NLP applications." (Sinan Ozdemir, "Quick Start Guide to Large Language Models: Strategies and Best Practices for Using ChatGPT and Other LLMs", 2024)

"The idea behind transfer learning is that the pre-trained model has already learned a lot of information about the language and relationships between words, and this information can be used as a starting point to improve performance on a new task. Transfer learning allows LLMs to be fine-tuned for specific tasks with much smaller amounts of task-specific data than would be required if the model were trained from scratch. This greatly reduces the amount of time and resources needed to train LLMs." (Sinan Ozdemir, "Quick Start Guide to Large Language Models: Strategies and Best Practices for Using ChatGPT and Other LLMs", 2024)

"Transfer learning is a technique used in machine learning to leverage the knowledge gained from one task to improve performance on another related task. Transfer learning for LLMs involves taking an LLM that has been pre-trained on one corpus of text data and then fine-tuning it for a specific 'downstream' task, such as text classification or text generation, by updating themodel’s parameters with task-specific data." (Sinan Ozdemir, "Quick Start Guide to Large Language Models: Strategies and Best Practices for Using ChatGPT and Other LLMs", 2024)

"Agentic intelligence feels incredibly powerful in demos but breaks in production. Indeed, it is very fragile without solid infrastructure. Every day, I personally see tons of clever orchestrations around dumb prompt chains tied up in a brittle, underused LLMOps infrastructure. But building this infrastructure means acknowledging the costs: performance overhead, strict interface contracts, and state complexity, as well as a need for more LLMOps engineers to create the best practices, tooling, and frameworks to run these systems reliably, safely, and robustly." (Abi Aryan, "LLMOps: Managing Large Language Models in Production", 2025)

"As the tech industry moves from non-generative models to generative models, it is shifting away from feature engineering, or creating features to model the data and experimenting with different hyperparameters to optimize performance. Generative models, and specifically LLMs, do not require feature engineering. Today, the core requirements are usually prompt engineering or building a RAG pipeline - skills that lie within the domain of AI engineers." (Abi Aryan, "LLMOps: Managing Large Language Models in Production", 2025)

"In prompt engineering, we customize the prompts or questions we give the model to get more accurate or insightful responses. The way a prompt is structured has a massive impact on how well a model understands the task at hand and, ultimately, how well it performs. Given LLMs’ versatility, prompt engineering has become an important skill for getting the most out of these models across different domains and tasks. The key is to understand how different prompt structures lead to different model behaviors. There are various strategies - ranging from simple one-shot prompting to more complex techniques like chain-of-thought prompting - that can significantly improve the effectiveness of LLMs." (Abi Aryan, "LLMOps: Managing Large Language Models in Production", 2025)

"LLM-centric workloads change everything. Now the raw material is heterogeneous text, code, images, audio, and chat logs whose value depends on semantic richness - that is, the informational value of the content - rather than a rigid structure. Pipelines must tokenize, chunk, embed, and version this content; store it in vector indexes for similarity search; and apply filters for personally identifiable information, toxicity, and licensing constraints. Instead of ETL jobs, teams run continuous ingestion and reembedding loops so that RAG systems stay fresh, and they log every prompt–response pair so that the inputs and outputs can be evaluated and improve the future performance of this system. Data quality in this context is judged by grounding, factuality, and bias metrics - attributes that require automated red-teaming and humanin-the-loop (HITL) review rather than the data structure violation checks of the past." (Abi Aryan, "LLMOps: Managing Large Language Models in Production", 2025)

🖍️Mark Needham - Collected Quotes

"A random walk, in general, is sometimes described as being similar to how a drunk person traverses a city. They know what direction or end point they want to reach but may take a very circuitous route to get there. The algorithm starts at one node and somewhat randomly follows one of the relationships forward or backward to a neighbor node. It then does the same from that node and so on, until it reaches the set path length. ('We say somewhat randomly because the number of relationships a node has, and its neighbors have, influences the probability a node will be walked through.)'" (Mark Needham & Amy E Hodler, "Graph Algorithms: Practical Examples in Apache Spark and Neo4j", 2019)

"Adding graph features and context improves predictions, especially in situations where connections matter. [...] Unfortunately, many machine learning approaches today miss a lot of rich contextual information. This stems from ML’s reliance on input data built from tuples, leaving out a lot of predictive relationships and network data. Furthermore, contextual information is not always readily available or is too difficult to access and process. Even finding connections that are four or more hops away can be a challenge at scale for traditional methods. Using graphs, we can more easily reach and incorporate connected data." (Mark Needham & Amy E Hodler, "Graph Algorithms: Practical Examples in Apache Spark and Neo4j", 2019)

"At the most abstract level, graph analytics is applied to forecast behavior and prescribe action for dynamic groups. Doing this requires understanding the relationships and structure within the group. Graph algorithms accomplish this by examining the overall nature of networks through their connections. With this approach, you can understand the topology of connected systems and model their processes." (Mark Needham & Amy E Hodler, "Graph Algorithms: Practical Examples in Apache Spark and Neo4j", 2019)

"Betweenness Centrality makes the assumption that all communication between nodes happens along the shortest path and with the same frequency, which isn’t always the case in real life. Therefore, it doesn’t give us a perfect view of the most influential nodes in a graph, but rather a good representation." (Mark Needham & Amy E Hodler, "Graph Algorithms: Practical Examples in Apache Spark and Neo4j", 2019)

"Centrality algorithms are used to understand the roles of particular nodes in a graph and their impact on that network. They’re useful because they identify the most important nodes and help us understand group dynamics such as credibility, accessibility, the speed at which things spread, and bridges between groups." (Mark Needham & Amy E Hodler, "Graph Algorithms: Practical Examples in Apache Spark and Neo4j", 2019)

"Community formation is common in all types of networks, and identifying them is essential for evaluating group behavior and emergent phenomena. The general prin‐ ciple in finding communities is that its members will have more relationships within the group than with nodes outside their group. Identifying these related sets reveals clusters of nodes, isolated groups, and network structure. This information helps infer similar behavior or preferences of peer groups, estimate resiliency, find nested relationships, and prepare data for other analyses. Community detection algorithms are also commonly used to produce network visualization for general inspection." (Mark Needham & Amy E Hodler, "Graph Algorithms: Practical Examples in Apache Spark and Neo4j", 2019)

"Feature extraction is a way to distill large volumes of data and attributes down to a set of representative descriptive attributes. The process derives numerical values (fea‐ tures) for distinctive characteristics or patterns in input data so that we can differenti‐ ate categories in other data. It’s used when data is difficult for a model to analyze directly - perhaps because of size, format, or the need for incidental comparisons. Feature selection is the process of determining the subset of extracted features that are most important or influential to a target goal. It’s used to surface predictive importance as well as for efficiency." (Mark Needham & Amy E Hodler, "Graph Algorithms: Practical Examples in Apache Spark and Neo4j", 2019)

"Graph algorithms provide one of the most potent approaches to analyzing connected data because their mathematical calculations are specifically built to operate on relationships. They describe steps to be taken to process a graph to discover its general qualities or specific quantities. Based on the mathematics of graph theory, graph algo‐ rithms use the relationships between nodes to infer the organization and dynamics of complex systems. Network scientists use these algorithms to uncover hidden infomation, test hypotheses, and make predictions about behavior." (Mark Needham & Amy E Hodler, "Graph Algorithms: Practical Examples in Apache Spark and Neo4j", 2019)

"Graph embedding is the representation of the nodes and relationships in a graph asfeature vectors. [...] Graph embedding uses graph data slightly differently than in connected feature extraction. It enables us to represent entire graphs, or subsets of graph data, in a numerical format ready for machine learning tasks. This is especially useful for unsu‐pervised learning, where the data is not categorized because it pulls in more contextual information through relationships. Graph embedding is also useful for data exploration, computing similarity between entities, and reducing dimensionality to aid in statistical analysis." (Mark Needham & Amy E Hodler, "Graph Algorithms: Practical Examples in Apache Spark and Neo4j", 2019)

"In classic graph theory, an acyclic graph that is undirected is called a tree. In computer science, trees can also be directed. A more inclusive definition would be a graph where any two nodes are connected by only one path. Trees are significant for understanding graph structures and many algorithms. They play a key role in designing networks, data structures, and search optimizations to improve categorization or organizational hierarchies." (Mark Needham & Amy E Hodler, "Graph Algorithms: Practical Examples in Apache Spark and Neo4j", 2019)

"Keep in mind that centrality measures represent the importance of a node in comparison to other nodes. Centrality is a ranking of the potential impact of nodes, not a measure of actual impact. For example, you might identify the two people with the highest cen‐ trality in a network, but perhaps policies or cultural norms are in play that actually shift influence to others. Quantifying actual impact is an active research area to develop additional influence metrics." (Mark Needham & Amy E Hodler, "Graph Algorithms: Practical Examples in Apache Spark and Neo4j", 2019)

"Machine learning is not artificial intelligence (AI), but a method for achieving AI. ML uses algorithms to train software through specific examples and progressive improvements based on expected outcome - without explicit programming of how to accomplish these better results. Training involves providing a lot of data to a model and enabling it to learn how to process and incorporate that information." (Mark Needham & Amy E Hodler, "Graph Algorithms: Practical Examples in Apache Spark and Neo4j", 2019)

"Make it a habit to run Connected Components to test whether a graph is connected as a preparatory step for general graph analysis. Performing this quick test can avoid accidentally running algorithms on only one disconnected component of a graph and getting incorrect results." (Mark Needham & Amy E Hodler, "Graph Algorithms: Practical Examples in Apache Spark and Neo4j", 2019)

"Sometimes the most important cog in the system is not the one with the most overt power or the highest status. Sometimes it’s the middlemen that connect groups or the brokers who the most control over resources or the flow of information. Betweenness Centrality is a way of detecting the amount of influence a node has over the flow of information or resources in a graph. It is typically used to find nodes that serve as a bridge from one part of a graph to another." (Mark Needham & Amy E Hodler, "Graph Algorithms: Practical Examples in Apache Spark and Neo4j", 2019)

"Putting together the right mix of features can increase accuracy because it fundamentally influences how our models learn. Because even modest improvements can make a significant difference, our focus in this chapter is on connected features. Connected features are features extracted from the structure of the data. These features can be derived from graph-local queries based on parts of the graph surrounding a node, or graph-global queries that use graph algorithms to identify predictive elements within data based on relationships for connected feature extraction." (Mark Needham & Amy E Hodler, "Graph Algorithms: Practical Examples in Apache Spark and Neo4j", 2019)

"Use Degree Centrality if you’re attempting to analyze influence by looking at the number of incoming and outgoing relationships, or find the “popularity” of individual nodes. It works well when you’re concerned with immediate connectedness or near-term probabilities. However, Degree Centrality is also applied to global analysis when you want to evaluate the minimum degree, maximum degree, mean degree, and standard deviation across the entire graph." (Mark Needham & Amy E Hodler, "Graph Algorithms: Practical Examples in Apache Spark and Neo4j", 2019)

"When using community detection algorithms, be conscious of the density of the relationships. If the graph is very dense, you may end up with all nodes congregating in one or just a few clusters. You can counteract this by filtering by degree, relationship weights, or similarity metrics. On the other hand, if the graph is too sparse with few connected nodes, you may end up with each node in its own cluster. In this case, try to incorporate additional relationship types that carry more relevant information." (Mark Needham & Amy E Hodler, "Graph Algorithms: Practical Examples in Apache Spark and Neo4j", 2019)

"Without peripheral and related information, solutions that attempt to predict behav‐ ior or make recommendations for varying circumstances require more exhaustive training and prescriptive rules. This is partly why AI is good at specific, well-defined tasks, but struggles with ambiguity. Graph-enhanced ML can help fill in that missing contextual information that is so important for better decisions." (Mark Needham & Amy E Hodler, "Graph Algorithms: Practical Examples in Apache Spark and Neo4j", 2019)