25 May 2018

Data Science: Multilayer Perceptron (Definitions)

"A neural net composed of three or more slabs (and therefore two or more layers of weighted connection paths); such nets are capable of solving more difficult problems than are single layer nets. They are often trained by backpropagation." (Laurene V Fausett, "Fundamentals of Neural Networks: Architectures, Algorithms, and Applications", 1994)

"A fully connected feedforward NN with at least one hidden layer that is trained using back-propagation algorithmic techniques." (Ioannis Papaioannou et al, "A Survey on Neural Networks in Automated Negotiations", Encyclopedia of Artificial Intelligence, 2009)

"A kind of feed-forward neural network which has at least one hidden layer of neurons." (Fernando Mateo et al, "A 2D Positioning Application in PET Using ANNs", Encyclopedia of Artificial Intelligence, 2009)

"A neural network that has one or more hidden layers, each of which has a linear combination function and executes a nonlinear activation function on the input to that layer." (Robert Nisbet et al, "Handbook of statistical analysis and data mining applications", 2009)

"It has a layered architecture consisting of input, hidden and output layers. Each layer consists of a number of perceptrons." (Siddhartha Bhattacharjee et al, "Quantum Backpropagation Neural Network Approach for Modeling of Phenol Adsorption from Aqueous Solution by Orange Peel Ash", 2013)

"A type of neural network. The MLP is the most common, and arguably the simplest, neural network used for classification." (Meta S Brown, "Data Mining For Dummies", 2014)

"An artificial neural network model with feed forward architecture that maps sets of input data onto a set of desired outputs iteratively, through the process of learning. A MLP consists of an input layer of neurons, one or more hidden layers of neurons and an output layer of neurons, where each layer is fully connected to the next layer." (Eitan Gross, "Stochastic Neural Network Classifiers", 2015) 

"an important class of ANN that typically consists of the input layer, one or more hidden layers of computation nodes, and an output layer. The input signal propagates through the network in a forward direction, on a layer-by-layer basis." (Pablo Escandell-Montero et al,"Artificial Neural Networks in Physical Therapy", 2015)

"Arguably the most popular artificial neural network model. It is usually composed by three or four layers of units. Each unit is fully connected to the units of the previous layer. Learning is customarily performed via the backpropagation rule." (D T Pham & M Castellani, "The Bees Algorithm as a Biologically Inspired Optimisation Method", 2015)

"Is an ANN type that requires a reference to learn patterns. It is trained using (error) back propagation algorithm." (Kandarpa K Sarma, "Learning Aided Digital Image Compression Technique for Medical Application", 2016)

"MLP is a feed forward neural network with one or more layers between input and output layer and are used to solve non-linearly separable problems. MLPs are trained using the back propagation algorithm. MLPs are widely used in pattern classification, recognition, prediction, etc." (Mridusmita Sharma & Kandarpa K Sarma, "Soft-Computational Techniques and Spectro-Temporal Features for Telephonic Speech Recognition: An Overview and Review of Current State of the Art", 2016)

Data Science: Neuron (Definitions)

[Chaotic neuron:] "An artificial neuron whose output is calculated with the use of a chaotic output function." (Nikola K Kasabov, "Foundations of Neural Networks, Fuzzy Systems, and Knowledge Engineering", 1996)

[Oscillatory neuron:] "An artificial neuron built up of two elements (or two groups of elements), one of them being excitatory and the other inhibitory. Its functioning is described as oscillation, characterized by three parameters: frequency; phase; amplitude." (Nikola K Kasabov, "Foundations of Neural Networks, Fuzzy Systems, and Knowledge Engineering", 1996)

"A nerve cell in the physiological nervous system." (Guido J Deboeck and Teuvo Kohonen, "Visual explorations in finance with self-organizing maps", 2000)

[Hidden neuron:] "Usually a nonlinear (or linear) processing element with no direct connections to either inputs or outputs. It often provides the learning capacity of the neural network." (Guido Deboeck & Teuvo Kohonen (Eds), "Visual Explorations in Finance with Self-Organizing Maps", 2000)

"any of the numerous types of specialized cell in the brain or other nervous systems that transmit and process neural signals. The nodes of artificial neural networks are also called neurons." (Teuvo Kohonen, "Self-Organizing Maps 3rd Ed.", 2001)

"A single processing element in a neural network. The most common form of neuron has two basic parts: a summation function that receives inputs and a transfer function that processes inputs and passes the processed values to the next layer of neurons. If the neuron is in the last layer of the network, the output is the final estimate of the dependent variable for that input vector or case." (David Scarborough & Mark J Somers, "Neural Networks in Organizational Research: Applying Pattern Recognition to the Analysis of Organizational Behavior", 2006)

"the elementary processing unit that composes an ANN." (Pablo Escandell-Montero et al, "Artificial Neural Networks in Physical Therapy", 2015)

"A neuron takes a number of input values (or activations) as input and maps these values to a single output activation. This mapping is typically implemented by applying a multi-input linear-regression function to the inputs and then pushing the result of this regression function through a nonlinear activation function, such as the logistic or tanh function." (John D Kelleher & Brendan Tierney, "Data science", 2018)

"Specialized brain cell that integrates inputs from other neurons and sends outputs to other neurons." (Terrence J Sejnowski, "The Deep Learning Revolution", 2018)

"A unit in a neural net whose function (such as y = tanh(w.dot(x))) takes multiple inputs and outputs a single scalar value. This value is usually the weights for that neuron (w or wi) multiplied by all the input signals (x or xi) and summed with a bias weight (w0) before applying an activation function like tanh. A neuron always outputs a scalar value, which is sent to the inputs of any additional hidden or output neurons in the network. If a neuron implements a much more complicated activation function than that, like the enhancements that were made to recurrent neurons to create an LSTM, it is usually called a unit, for example, an LSTM unit." (Hobson Lane et al, "Natural Language Processing in Action: Understanding, analyzing, and generating text with Python", 2019)

"An artificial neuron is a model of a neuron present in an animal brain that is perceived as a mathematical function." (Hari K Kondaveeti et al, "Deep Learning Applications in Agriculture: The Role of Deep Learning in Smart Agriculture", 2021)


24 May 2018

Data Science: Pattern Recognition (Definitions)

"The categorization of patterns in some domain into meaningful classes. A pattern usually has the form of a vector of measurement values." (Guido Deboeck & Teuvo Kohonen (Eds), "Visual Explorations in Finance with Self-Organizing Maps 2nd Ed.", 2000)

"in the most general sense the same as artificial perception." (Teuvo Kohonen, "Self-Organizing Maps" 3rd Ed., 2001)

"The operation and design of systems that recognize patterns in data." (Craig F Smith & H Peter Alesso, "Thinking on the Web: Berners-Lee, Gödel and Turing", 2008)

"Research area that enclose the development of methods and automatized techniques for identification and classification of samples in specific groups, in accordance with representative characteristics." (Paulo E Ambrósio, "Artificial Intelligence in Computer-Aided Diagnosis",  Encyclopedia of Artificial Intelligence, 2009)

"The process of identifying patterns in data via algorithms to make predictions within a subject area." (Jason Williamson, Getting a Big Data Job For Dummies, 2015)

"A branch of machine learning that recognizes and separates the patterns of one class from the other." (Mridusmita Sharma & Kandarpa K Sarma, "Soft-Computational Techniques and Spectro-Temporal Features for Telephonic Speech Recognition: An Overview and Review of Current State of the Art", 2016)

"A pattern is a particular configuration of data; for example, ‘A’ is a composition of three strokes. Pattern recognition is the detection of such patterns." (Ethem Alpaydın, "Machine learning : the new AI", 2016)

"Pattern Recognition in the discipline which tries to find the classes in the datasets of the various applications and it is the major building block of artificially intelligent systems." (Vandana M Ladwani, "Support Vector Machines and Applications", 2017)

"identifying patterns in data via algorithms to make predictions of new data coming from the same source." (Analytics Insight)

23 May 2018

Data Science: Markov Process (Definitions)

"stochastic process in which the new state of a system depends on the previous state only (or more generally, on a finite set of previous states)." (Teuvo Kohonen, "Self-Organizing Maps" 3rd Ed., 2001)

"A stochastic process in which the transition probabilities can be estimated on the basis of first order data. Such a process is also stationary in that probability estimates do not change across the sample (generally across time)." (W David Penniman,"Historic Perspective of Log Analysis", 2009)

"Stochastic process in which the new state of a system depends on the previous state or a finite set of previous states." (Patrick Rousset & Jean-Francois Giret, "A Longitudinal Analysis of Labour Market Data with SOM" Encyclopedia of Artificial Intelligence, 2009)

"A stochastic process where the probabilities of the events depend on the previous event only." (Michael M Richter, "Business Processes, Dynamic Contexts, Learning", 2014)

"A Markov chain (or Markov process) is a system containing a finite number of distinct states S1,S2,…,Sn on which steps are performed such that: (1) At any time, each element of the system resides in exactly one of the states. (2) At each step in the process, elements in the system can move from one state to another. (3) The probabilities of moving from state to state are fixed - that is, they are the same at each step in the process." (Stephen Andrilli & David Hecker, [in [Elementary Linear Algebra] 5th Ed.), 2016)

[hidden Markov model:] "A hidden Markov model is a technique for modeling sequences using a hidden state that only uses the previous part of the sequence." (Alex Thomas, "Natural Language Processing with Spark NLP", 2020)

[Markov decision process:] "A stochastic dynamic program, whereby for each policy the resulting state variables comprise a Markov process (a stochastic process with the property that the conditional probability of a transition to any state depends only on the current state, and not on previous states)." (Mathematical Programming Glossary)

22 May 2018

Data Science: Time Series (Definitions)

"It is composed of a sequence of values, where each value corresponds to a time instance. The length remains constant." (Maria Kontaki et al, "Similarity Search in Time Series",  2009)

"a time series is a sequence of data points, measured typically at successive times, spaced at time intervals." (Yong Yu et al, "Applications of Evolutionary Neural Networks for Sales Forecasting of Fashionable Products", 2010)

"A sequence of numerical values of a variable obtained at some regular/uniform intervals of time or at non uniform intervals of time." (Mofazzal H Khondekar et al, "Soft Computing Based Statistical Time Series Analysis, Characterization of Chaos Theory, and Theory of Fractals", 2013)

"A series of values of a quantity obtained at successive times, often with equal intervals between them." (Dima Alberg & Zohar Laslo, "Segmenting Big Data Time Series Stream Data", 2014) 

"An ordered sequence of values that correspond to a variable that is typically sampled at a uniform sampling rate. Time series prediction is intended to make estimations about the future values of the series." (Fernando Mateo et al, "Forecasting Techniques for Energy Optimization in Buildings", 2015)

"A sequence of data points consisting of consecutive measurements that are made over a time interval." (Vasileios Zois, "Querying of Time Series for Big Data Analytics", 2016)

"A series of values of a quantity obtained at successive times, often with equal intervals between them." (Dima Alberg, "Big Data Time Series Stream Data Segmentation Methods", Encyclopedia of Information Science and Technology, 2018)

"A series of data points indexed (or listed or graphed) in time order. Most commonly, a time series is a sequence taken at successive equally spaced points in time." (Gurpreet Kaur & Akriti Gupta, "India-BIMSTEC Bilateral Trade Activities: A Gravity Model Approach", 2020)

"Time series is a series of data points that are listed in time order." (Siyu Shi, "Introduction to Python and Its Statistical Applications", 2020)

"A set of successive observations collected generally at the same interval, named period." (Oumayma Bounouh et al, "Investigating the Pixel Quality Influence on Forecasting Vegetation Change Dynamics: Application Case of Tunisian Olive Sites", 2021)

Data Science: Recurrent Neural Network (Definitions)

"A neural net with feedback connections, such as a BAM, Hopfield net, Boltzmann machine, or recurrent backpropagation net. In contrast, the signal in a feedforward neural net passes from the input units (through any hidden units) to the output units." (Laurene V Fausett, "Fundamentals of Neural Networks: Architectures, Algorithms, and Applications", 1994)

"A neural network topology where the units are connected so that inputs signals flow back and forth between the neural processing units until the neural network settles down. The outputs are then read from the output units." (Joseph P Bigus, "Data Mining with Neural Networks: Solving Business Problems from Application Development to Decision Support", 1996)

"Networks with feedback connections from neurons in one layer to neurons in a previous layer." (Nikola K Kasabov, "Foundations of Neural Networks, Fuzzy Systems, and Knowledge Engineering", 1996)

"RNN topology involves backward links from output to the input and hidden layers." (Siddhartha Bhattacharjee et al, "Quantum Backpropagation Neural Network Approach for Modeling of Phenol Adsorption from Aqueous Solution by Orange Peel Ash", 2013)

"Neural network whose feedback connections allow signals to circulate within it." (Terrence J Sejnowski, "The Deep Learning Revolution", 2018)

"An RNN is a special kind of neural network used for modeling sequential data." (Alex Thomas, "Natural Language Processing with Spark NLP", 2020)

"A recurrent neural network (RNN) is a class of artificial neural networks where connections between nodes form a directed graph along a temporal sequence. This allows it to exhibit temporal dynamic behavior." (Udit Singhania & B. K. Tripathy, "Text-Based Image Retrieval Using Deep Learning", 2021)

"A RNN [Recurrent Neural Network] models sequential interactions through a hidden state, or memory. It can take up to N inputs and produce up to N outputs. For example, an input sequence may be a sentence with the outputs being the part-of-speech tag for each word (N-to-N). An input could be a sentence, and the output a sentiment classification of the sentence (N-to-1). An input could be a single image, and the output could be a sequence of words corresponding to the description of an image (1-to-N). At each time step, an RNN calculates a new hidden state ('memory') based on the current input and the previous hidden state. The 'recurrent' stems from the facts that at each step the same parameters are used and the network performs the same calculations based on different inputs." (Wild ML)

"Recurrent Neural Network (RNN) refers to a type of artificial neural network used to understand sequential information and predict follow-on probabilities. RNNs are widely used in natural language processing, with applications including language modeling and speech recognition." (Accenture)

21 May 2018

Data Science: Generative Adversarial Network (Definitions)

"A category of deep learning neural networks that are composed of two competitive neural networks together." (Dulani Meedeniya & Iresha Rubasinghe, "A Review of Supportive Computational Approaches for Neurological Disorder Identification", 2020) 

"A powerful machine learning technique made up of two learning systems that compete with each other in a game-like fashion. Features of the winning system are 'genetically' added to the loser along with random mutations. GANs teach themselves through this 'survival of the fittest' evolutionary model. They 'generate' new solutions through many, often millions, of generations." (Scott R Garrigan, "Frameworks for Integration of Future-Oriented Computational Thinking in K-12 Schools", 2020)

"An artificial intelligence process that includes a 'generator' that produces samples, and a 'discriminator' that differentiates between computer-generated samples and samples derived from 'real-world' sources." (Keram Malicki-Sanchez, "Out of Our Minds: Ontology and Embodied Media in a Post-Human Paradigm", 2020)

"Machine learning framework in which two neural networks compete against each other to win within a gaming environment using a supervised learning pattern." (Jose A R Pinheiro, "Contemporary Imagetics and Post-Images in Digital Media Art: Inspirational Artists and Current Trends (1948-2020)", 2020)

"It refers to a type of neural network that consists of a generative and a discriminative network that contest with each other especially in a game scenario. They are used to generate new data that are statistically similar to the training data." (Vijayaraghavan Varadharajan & J Rian Leevinson, "Next Generation of Intelligent Cities: Case Studies from Europe", 2021)

"A generative adversarial network, or GAN, is a deep neural network framework which is able to learn from a set of training data and generate new data with the same characteristics as the training data." (Thomas Wood)

20 May 2018

Data Science: Semi-supervised Learning (Definitions)

"machine learning technique that uses both labelled and unlabelled data for constructing the model." (Óscar Pérez & Manuel Sánchez-Montañés, "Class Prediction in Test Sets with Shifted Distributions", 2009)

"The set of learning algorithms in which the samples in training dataset are all unlabelled." (Jun Jiang & Horace H S Ip, "Active Learning with SVM, Encyclopedia of Artificial Intelligence", 2009) 

"Learning to label new data using both labeled training data plus unlabeled data." (Jesse Read & Albert Bifet, "Multi-Label Classification", 2014)

"A method of empirical concept learning from unlabeled data. The task is to build a model that finds groups of similar examples or that finds dependencies between attribute-value tuples." (Petr Berka, "Machine Learning", 2015)

"Combines the methodology of the supervised learning to process the labeled data with the unsupervised learning to compute the unlabeled data." (Nuno Pombo et al, "Machine Learning Approaches to Automated Medical Decision Support Systems", 2015)

"Estimation of the parameters of a model considering only un-labeled data and without the help of human experts." (Manuel Martín-Merino, "Semi-Supervised Dimension Reduction Techniques to Discover Term Relationships", 2015)

"In this category either the model is developed in such a way that either there are labels exist for all kind of observations or there is no label exist." (Neha Garg & Kamlesh Sharma, "Machine Learning in Text Analysis", 2020)

"It is a machine learning algorithm in which the machine learns from both labeled and unlabeled instances to build a model for predicting the class of unlabeled instances." (Gunjan Ansari et al, "Natural Language Processing in Online Reviews", 2021)

"Semi-supervised learning aims at labeling a set of unlabelled data with the help of a small set of labeled data." (Hari K Kondaveeti et al, "Deep Learning Applications in Agriculture: The Role of Deep Learning in Smart Agriculture", 2021)

"The semi-supervised learning combines both supervised and unsupervised learning algorithms." (M Govindarajan, "Big Data Mining Algorithms", 2021)

19 May 2018

Data Science: Perceptron (Definitions)

"the term is often used to refer to a single layer pattern classification network with linear threshold units" (Laurene V Fausett, "Fundamentals of Neural Networks: Architectures, Algorithms, and Applications", 1994)

"adaptive element for multilayer feedforward networks introduced by Rosenblatt" (Teuvo Kohonen, "Self-Organizing Maps" 3rd Ed., 2001)

"An early theoretical model of the neuron developed by Rosenblatt (1958) that was the first to incorporate a learning rule. The term is also used as a generic label for all trained feed-forward networks, which is often referred to collectively as multilayer perceptron networks." (David Scarborough & Mark J Somers, "Neural Networks in Organizational Research: Applying Pattern Recognition to the Analysis of Organizational Behavior", 2006)

"A type of binary classifier that maps its inputs (a vector of real type) to an output value (a scalar real type). The perceptron may be considered as the simplest model of feed-forward neural network, as the inputs directly feeding the output units through weighted connections." )Crescenzio Gallo, "Artificial Neural Networks Tutorial", 2015)

"A perceptron is a type of a neural network organized into layers where each layer receives connections from units in the previous layer and feeds its output to the units of the layer that follow." (Ethem Alpaydın, "Machine learning : the new AI", 2016)

"Perceptron is a learning algorithm which is used to learn the decision boundary for linearly separable data." Vandana M Ladwani, "Support Vector Machines and Applications", 2017)

"A simple neural network model consisting of one unit and inputs with variable weights that can be trained to classify inputs into categories." (Terrence J Sejnowski, "The Deep Learning Revolution", 2018)

"The simplest form of artificial neural network, a basic operational unit which employs supervised learning. It is used to classify data into two classes." (Gaetano B Ronsivalle & Arianna Boldi, "Artificial Intelligence Applied: Six Actual Projects in Big Organizations", 2019)

"A perceptron is a single-layer neural network. It includes input values, weights and bias, net sum, and an activation function." (Prisilla Jayanthi & Muralikrishna Iyyanki, "Deep Learning Techniques for Prediction, Detection, and Segmentation of Brain Tumors", 2020)

"The basic unit of a neural network that encodes inputs from neurons of the previous layer using a vector of weights or parameters associated with the connections between perceptrons." Mário P Véstias, "Deep Learning on Edge: Challenges and Trends", 2020)

"these are machine learning algorithms that undertake supervised learning of functions called binary classifiers which decide whether or not an input, usually identified with a vector of numbers, belongs to a particular class." (Hari Kishan Kondaveeti et al, "Deep Learning Applications in Agriculture: The Role of Deep Learning in Smart Agriculture", 2021)

Data Science: Convolutional Neural Network (Definitions)

"A multi layer neural network similar to artificial neural networks only differs in its architecture and mainly built to recognize visual patterns from image pixels." (Nishu Garg et al, "An Insight Into Deep Learning Architectures, Latent Query Features", 2018)

"In machine learning, a convolutional neural network is a class of deep, feed-forward artificial neural networks that has successfully been applied to analyzing visual imagery. CNNs use a variation of multilayer perceptrons designed to require minimal preprocessing. They are also known as shift invariant or space invariant artificial neural networks (SIANN), based on their shared-weights architecture and translation invariance characteristics." (V E Jayanthi, "Automatic Detection of Tumor and Bleed in Magnetic Resonance Brain Images", 2018)

"A special type of feed-forward neural network optimized for image data processing. The key features of CNN architecture include sharing weights, using pooling layers, implementing deep structures with multiple hidden layers." (Lyudmila N. Tuzova et al, "Teeth and Landmarks Detection and Classification Based on Deep Neural Networks", 2019)

"A type of artificial neural networks, which uses a set of filters with tunable (learnable) parameters to extract local features from the input data." (Sergei Savin & Aleksei Ivakhnenko, "Enhanced Footsteps Generation Method for Walking Robots Based on Convolutional Neural Networks", 2019) 

"A convolutional neural network (CNN) is a type of artificial neural network used in image recognition and processing that is specifically designed to process pixel data by means of learnable filters." (Loris Nanni et al, "Digital Recognition of Breast Cancer Using TakhisisNet: An Innovative Multi-Head Convolutional Neural Network for Classifying Breast Ultrasonic Images", 2020)

"A convolutional neural network (CNN) is a type of artificial neural network used in image recognition and processing that is specifically designed to process pixel data. CNNs are powerful image processing, artificial intelligence (AI) that use deep learning to perform both generative and descriptive tasks, often using machine vision that includes image and video recognition, along with recommender systems and natural language processing (NLP)." (Mohammad F Hashmi et al, "Subjective and Objective Assessment for Variation of Plant Nitrogen Content to Air Pollutants Using Machine Intelligence", 2020)

"A neural network with a convolutional layer which does the mathematical operation of convolution in addition to the other layers of deep neural network." (S Kayalvizhi & D Thenmozhi, "Deep Learning Approach for Extracting Catch Phrases from Legal Documents", 2020)

"A special type of neural networks used popularly to analyze photography and imagery." (Murad Al Shibli, "Hybrid Artificially Intelligent Multi-Layer Blockchain and Bitcoin Cryptology", 2020)

"In deep learning, a convolutional neural network is a class of deep neural networks, most commonly applied to analyzing visual imagery. CNNs use a variation of multilayer perceptrons designed to require minimal preprocessing." (R Murugan, "Implementation of Deep Learning Neural Network for Retinal Images", 2020)

"A class of deep neural networks applied to image processing where some of the layers apply convolutions to input data." (Mário P Véstias, "Convolutional Neural Network", 2021)

"A convolution neural network is a kind of ANN used in image recognition and processing of image data." (M Srikanth Yadav & R Kalpana, "A Survey on Network Intrusion Detection Using Deep Generative Networks for Cyber-Physical Systems", 2021)

"A multi-layer neural network similar to artificial neural networks only differs in its architecture and mainly built to recognize visual patterns from image pixels." (Udit Singhania & B K Tripathy, "Text-Based Image Retrieval Using Deep Learning", 2021) 

"A type of deep learning algorithm commonly applied in analyzing image inputs." (Jinnie Shin et al, "Automated Essay Scoring Using Deep Learning Algorithms", 2021)

"It is a class of deep neural networks, most commonly applied to analyzing visual imagery." (Sercan Demirci et al, "Detection of Diabetic Retinopathy With Mobile Application Using Deep Learning", 2021)

"They are a class of deep neural networks that are generally used to analyze image data. They use convolution instead of simple matrix multiplication in a few layers of the network. They have shared weights architecture and have translation invariant characteristics." Vijayaraghavan Varadharajan & J Rian Leevinson, "Next Generation of Intelligent Cities: Case Studies from Europe", 2021) 

18 May 2018

Data Science: Boltzmann Machine (Definitions)

[Boltzmann machine (with learning):] "A net that adjusts its weights so that the equilibrium configuration of the net will solve a given problem, such as an encoder problem" (David H Ackley et al, "A learning algorithm for boltzmann machines", Cognitive Science Vol. 9 (1), 1985)

[Boltzmann machine (without learning):] "A class of neural networks used for solving constrained optimization problems. In a typical Boltzmann machine, the weights are fixed to represent the constraints of the problem and the function to be optimized. The net seeks the solution by changing the activations (either 1 or 0) of the units based on a probability distribution and the effect that the change would have on the energy function or consensus function for the net." (David H Ackley et al, "A learning algorithm for boltzmann machines", Cognitive Science Vol. 9 (1), 1985)

"neural-network model otherwise similar to a Hopfield network but having symmetric interconnects and stochastic processing elements. The input-output relation is optimized by adjusting the bistable values of its internal state variables one at a time, relating to a thermodynamically inspired rule, to reach a global optimum." (Teuvo Kohonen, "Self-Organizing Maps 3rd" Ed., 2001)

"A neural network model consisting of interacting binary units in which the probability of a unit being in the active state depends on its integrated synaptic inputs." (Terrence J Sejnowski, "The Deep Learning Revolution", 2018)

"An unsupervised network that maximizes the product of probabilities assigned to the elements of the training set." (Mário P Véstias, "Deep Learning on Edge: Challenges and Trends", 2020)

"Restricted Boltzmann machine (RBM) is an undirected graphical model that falls under deep learning algorithms. It plays an important role in dimensionality reduction, classification and regression. RBM is the basic block of Deep-Belief Networks. It is a shallow, two-layer neural networks. The first layer of the RBM is called the visible or input layer while the second is the hidden layer. In RBM the interconnections between visible units and hidden units are established using symmetric weights." (S Abirami & P Chitra, "The Digital Twin Paradigm for Smarter Systems and Environments: The Industry Use Cases", Advances in Computers, 2020)

"A deep Boltzmann machine (DBM) is a type of binary pairwise Markov random field (undirected probabilistic graphical model) with multiple layers of hidden random variables." (Udit Singhania & B. K. Tripathy, "Text-Based Image Retrieval Using Deep Learning",  2021) 

"A Boltzmann machine is a neural network of symmetrically connected nodes that make their own decisions whether to activate. Boltzmann machines use a straightforward stochastic learning algorithm to discover “interesting” features that represent complex patterns in the database." (DeepAI) [source]

"Boltzmann Machines is a type of neural network model that was inspired by the physical process of thermodynamics and statistical mechanics. [...] Full Boltzmann machines are impractical to train, which is one of the reasons why a limited form, called the restricted Boltzmann machine, is used." (Accenture)

"RBMs [Restricted Boltzmann Machines] are a type of probabilistic graphical model that can be interpreted as a stochastic artificial neural network. RBNs learn a representation of the data in an unsupervised manner. An RBN consists of visible and hidden layer, and connections between binary neurons in each of these layers. RBNs can be efficiently trained using Contrastive Divergence, an approximation of gradient descent." (Wild ML)

Data Science: Natural Language Processing (Definitions)

"Using software to 'understand' the meaning contained within texts. Everyday speech is broken down into patterns. Typically, these systems employ syntactic analysis to infer the semantic meaning embedded in documents. NLP identifies patterns in sample texts and makes predictions about unseen texts." (Craig F Smith & H Peter Alesso, "Thinking on the Web: Berners-Lee, Gödel and Turing", 2008)

"Use of computers to interpret and manipulate words as part of a language." (Dougal Hutchison, "Automated Essay Scoring Systems", 2009)

"It is a subfield of Computational Linguistics (i.e. the field that researches linguistics phenomena that occur in digital data), whose focus is on how to build automatic systems able to interpret/generate information in natural language." (Diana Pérez-Marín et al, "Adaptive Computer Assisted Assessment", 2010)

"the notion that the context of text can be inferred from the text itself." (Daniel Linstedt & W H Inmon, "Data Architecture: A Primer for the Data Scientist", 2014)

"An area of computer science involved with the computational study of human languages." (Jason Williamson, "Getting a Big Data Job For Dummies", 2015)

"Similarly to text mining, NLP is a multidisciplinary research field of computer science, artificial intelligence, and linguistics. However, it mainly focuses on the interaction between computers and human languages." (Hamid R Arabnia et al, "Application of Big Data for National Security", 2015)

"Natural Language Processing is prevalently used to analyse the text or speech in order to make machine understand the words like human." (Anumeera Balamurali & Balamurali Ananthanarayanan,"Develop a Neural Model to Score Bigram of Words Using Bag-of-Words Model for Sentiment Analysis", 2020)

 "Natural language processing is the ability of computer program to understand human language as it is spoken or handwritten." (Neha Garg & Kamlesh Sharma, "Machine Learning in Text Analysis", 2020)

"NLP is a field of computer science and linguistics focused on techniques and algorithms for processing data, continuing natural language." (Alex Thomas, "Natural Language Processing with Spark NLP", 2020)

"NLP is a Linguistic approach to interact with human language and computer. This field comes under Artificial Intelligence and Computer Science." (Sayani Ghosal & Amita Jain, "Research Journey of Hate Content Detection From Cyberspace", 2021)

"a field of computer science involved with interactions between computers and human languages." (Analytics Insight)

"is a field of computer science, with the goal to understand or generate human languages, either in text or speech form. There are two primary sub fields of NLP, Natural Language Understanding (NLU), and Natural Language Generation (NLG)." (Accenture)

17 May 2018

Data Science: Learning (Definitions)

"Procedures for modifying the weights on the connection links in a neural net (also known as training algorithms, learning rules)." (Laurene V Fausett, "Fundamentals of Neural Networks: Architectures, Algorithms, and Applications", 1994)

"In the simplest form: self-adaptation at the processing element level. Weighted connections between processing elements or weights are adjusted to achieve specific results, eliminating the need for writing a specific algorithm for each problem. More generally: change of rules or behavior for a certain objective." (Guido J Deboeck and Teuvo Kohonen, "Visual explorations in finance with self-organizing maps", 2000)

"generic name for all behavioral changes that depend on experiences and improve the performance of a system. In a more restricted sense learning is identical with adaptation, especially selective modification of parameters of a system." (Teuvo Kohonen, "Self-Organizing Maps" 3rd Ed., 2001)

"A process whereby a training set of examples is used to generate a model that understands and generalizes the relationship between the descriptor variables and one or more response variables." (Glenn J Myatt, "Making Sense of Data: A Practical Guide to Exploratory Data Analysis and Data Mining", 2006)

"The process of automatically finding relations between inputs and outputs given examples of that relation." (Craig F Smith & H Peter Alesso, "Thinking on the Web: Berners-Lee, Gödel and Turing", 2008)

"An essential operation of acquiring, processing and storing information required by any intelligent system for evolution." (T R Gopalakrishnan Nair, "Cognitive Approaches for Intelligent Networks", 2015)

"Adaptation of synaptic weights of a neural network as training progresses, usually with the objective of minimizing a cost function." (Anand Parey & Amandeep S Ahuja, "Application of Artificial Intelligence to Gearbox Fault Diagnosis: A Review", 2016)

"Algorithm for changing the parameters of a function based on examples. Learning algorithms are said to be “supervised” when both inputs and desired outputs are given or “unsupervised” when only inputs are given. Reinforcement learning is a special case of a supervised learning algorithm when the only feedback is a reward for good performance." (Terrence J Sejnowski, "The Deep Learning Revolution", 2018)

"A phase in the machine learning methods that aggregates some information about the state actions for using in the future predictions of the events." (Derya Yiltas-Kaplan, "The Usage Analysis of Machine Learning Methods for Intrusion Detection in Software-Defined Networks", 2019)

Data Science: Type I Error (Definitions)

"Within a hypothesis test, a type I error is the error of incorrectly rejecting a null hypothesis when it is true." (Glenn J Myatt, "Making Sense of Data: A Practical Guide to Exploratory Data Analysis and Data Mining", 2006)

"A type of error used in hypothesis testing that arises when incorrectly rejecting the null hypothesis, although it is actually true. Thus, based on the test statistic, the final conclusion rejects the Null hypothesis, but in truth it should be accepted. Type I error equates to the alpha (α) or significance level, whereby the generally accepted default is 5%." (Lynne Hambleton, "Treasure Chest of Six Sigma Growth Methods, Tools, and Best Practices", 2007)

"A term that refers to incorrectly rejecting a null hypothesis. It is also sometimes termed a false positive. It is used when an outcome is incorrectly identified as having happened, such as when a customer is incorrectly identified as having committed fraud." (Evan Stubbs, "Delivering Business Analytics: Practical Guidelines for Best Practice", 2013)

"Rejection of the null hypothesis when it's true." (Geoff Cumming, "Understanding The New Statistics", 2013)

"Probability of rejecting the null hypothesis when the null hypothesis is true." (K  N Krishnaswamy et al, "Management Research Methodology: Integration of Principles, Methods and Techniques", 2016)

"Probability of rejecting the null hypothesis when it's true." (Geoff Cumming, "Understanding The New Statistics", 2013)

Data Science: Unsupervised Learning (Definitions)

"A means of modifying the weights of a neural net without specifying the desired output for any input patterns. Used in self-organizing neural nets for clustering data, extracting principal components, or curve fitting." (Laurene V Fausett, "Fundamentals of Neural Networks: Architectures, Algorithms, and Applications", 1994)

"Learning in which no teacher is used to show the correct response to a given input stimulus; the system must organize itself purely on the basis of the input stimuli it receives. Often synonymous with clustering." (Guido J Deboeck & Teuvo Kohonen, "Visual explorations in finance with self-organizing maps", 2000)

"learning without a priori knowledge about the classification of samples; learning without a teacher. Often the same as formation of clusters, where after these clusters can be labeled. Also optimal allocation of computing resources when only unlabeled, unclassified data are input." (Teuvo Kohonen, "Self-Organizing Maps 3rd Ed.", 2001)

"Analysis methods that do not use any data to guide the technique operations." (Glenn J Myatt, "Making Sense of Data: A Practical Guide to Exploratory Data Analysis and Data Mining", 2006)

"Learning techniques that group instances without a pre-specified dependent attribute. Clustering algorithms are usually unsupervised methods for grouping data sets." (Lluís Formiga & Francesc Alías, "GTM User Modeling for aIGA Weight Tuning in TTS Synthesis", Encyclopedia of Artificial Intelligence, 2009)

"Method of machine learning where a model is fit to observations. It is distinguished from supervised learning by the fact that there is no a priori output." (Soledad Delgado et al, "Growing Self-Organizing Maps for Data Analysis", Encyclopedia of Artificial Intelligence, 2009)

"The type of learning that occurs when algorithms adjust the weights in a neural network by reference to a training data set that includes input variables only. Unsupervised learning algorithms attempt to locate clusters in the input data." (Robert Nisbet et al, "Handbook of statistical analysis and data mining applications", 2009)

"Treats all variables the same way so as to determine the different classes based on diverse features observed in the collection of unlabeled data that encompass the sample set. It is assumed that the user is unaware of the classes due to the lack of information sufficiently available." (Nuno Pombo et al, "Machine Learning Approaches to Automated Medical Decision Support Systems", 2015)

"Unsupervised learning refers to a machine learning approach that uses inferential statistical modeling algorithms to discover rather than detect patterns or similarities in data. An unsupervised learning system can identify new patterns, instead of trying to match a set of patterns it encountered during training." (Judith S Hurwitz, "Cognitive Computing and Big Data Analytics", 2015)

"In unsupervised learning, the program gets the same items but has to come up with the categories on its own by discovering the underlying correlations between the items; that is why unsupervised learning is sometimes called statistical pattern recognition." (Robert J Glushko, "The Discipline of Organizing: Professional Edition, 4th Ed", 2016)

"A form of machine learning in which the goal is to identify regularities in the data. These regularities may include clusters of similar instances within the data or regularities between attributes. In contrast to supervised learning, in unsupervised learning no target attribute is defined in the data set." (John D Kelleher & Brendan Tierney, "Data science", 2018)

"Unsupervised learning identifies hidden patterns or intrinsic structures in the data. It is used to draw conclusions from datasets composed of labeled unacknowledged input data." (Soraya Sedkaoui, "Big Data Analytics for Entrepreneurial Success", 2018)

"Unsupervised learning or clustering is a way of discovering hidden structures in unlabeled data. Clustering algorithms aim to discover latent patterns in unlabeled data using features to organize instances into meaningfully dissimilar groups." (Benjamin Bengfort et al, "Applied Text Analysis with Python: Enabling Language-Aware Data Products with Machine Learning", 2018)

"A particular form of learning process that takes place without supervision and that affects the training of an artificial neural networks." (Gaetano B Ronsivalle & Arianna Boldi, "Artificial Intelligence Applied: Six Actual Projects in Big Organizations", 2019)

"In this learning, the model does not require labeled data for training. The model learns the nature of data and does predictions." (Aman Kamboj et al, "Ear Localizer: A Deep-Learning-Based Ear Localization Model for Side Face Images in the Wild", 2019)

"A class of machine learning techniques designed to identify features and patterns in data. There is no mapping function to be learned or output values to be achieved. Cluster analysis is an example of unsupervised learning." (Timofei Bogomolov et al, "Identifying Patterns in Fresh Produce Purchases: The Application of Machine Learning Techniques", 2020)

"Unsupervised algorithms mean that a program is provided with some collection of data, with no predetermined dataset being available." (Neha Garg & Kamlesh Sharma, "Machine Learning in Text Analysis", 2020)

"A machine learning technique that involves providing a machine with data that is not labeled, instead allowing for the machine to learn by association." (Sujata Ramnarayan, "Marketing and Artificial Intelligence: Personalization at Scale", 2021)

"Unsupervised Learning aims at inferring the given unlabelled data using a different type of structures present in the data points." (Hari K Kondaveeti et al, "Deep Learning Applications in Agriculture: The Role of Deep Learning in Smart Agriculture", 2021)

"Unsupervised Learning is a type of machine learning in which the algorithm does not need the data with pre-defined labels. Unsupervised machine learning instead categorizes entries within datasets by examining similarities or anomalies and then grouping different entries accordingly." (Accenture)

16 May 2018

Data Science: Training Set/Dataset (Definitions)

"set of data used as inputs in an adaptive process that teaches a neural network." (Teuvo Kohonen, "Self-Organizing Maps" 3rd Ed., 2001)

"A set of observations that are used in creating a prediction model." (Glenn J Myatt, "Making Sense of Data: A Practical Guide to Exploratory Data Analysis and Data Mining", 2006)

"the training set is composed by all labelled examples that are provided for constructing a classifier. The test set is composed by the new unlabelled patterns whose classes should be predicted by the classifier." (Óscar Pérez & Manuel Sánchez-Montañés, "Class Prediction in Test Sets with Shifted Distributions", 2009)

"A collection of data whose purpose is to be analyzed to discover patterns that can then be applied to other data sets." (DAMA International, "The DAMA Dictionary of Data Management", 2011)

"A training set for supervised learning is taken from the labeled instances. The remaining instances are used for validation." (Robert J Glushko, "The Discipline of Organizing: Professional Edition" 4th Ed., 2016)

"A set of known and predictable data used to train a data mining model." (Microsoft, "SQL Server 2012 Glossary", 2012)

"In data mining, a sample of data used at each iteration of the training process to evaluate the model fit." (Meta S Brown, "Data Mining For Dummies", 2014)

"Training Data is the data used to train a machine learning algorithm. Generally, data in machine learning is divided into three datasets: training, validation and testing data. In general, the more accurate and comprehensive training data is, the better the algorithm or classifier will perform." (Accenture)

Data Science: Type II Error (Definitions)

"Within a hypothesis test, a type II error is the error of incorrectly not rejecting a null hypothesis when it should be rejected." (Glenn J Myatt, "Making Sense of Data: A Practical Guide to Exploratory Data Analysis and Data Mining", 2006)

"A type of error used in hypothesis testing that occurs when the test decision incorrectly “accepts” the null hypothesis. Based on the test statistic, the final decision fails to reject the Null when it is actually false. Type II error also is called 'beta' (β), and the default is typically set at 20%." (Lynne Hambleton, "Treasure Chest of Six Sigma Growth Methods, Tools, and Best Practices", 2007)

"A term that refers to failing to reject a null hypothesis when it is false. It is also sometimes termed a false negative and used when an outcome is incorrectly identified as not having happened, such as when a customer has committed fraud but has not been accurately identified." (Evan Stubbs, "Delivering Business Analytics: Practical Guidelines for Best Practice", 2013)

"Nonrejection of the null hypothesis when it's false." (Geoff Cumming, "Understanding The New Statistics", 2013)

"When the system accepts impostors who should be rejected (false acceptance rate)." (Adam Gordon, "Official (ISC)2 Guide to the CISSP CBK" 4th Ed., 2015)

"Probability of not rejecting the null hypothesis when the null hypothesis is false." (K  N Krishnaswamy et al, "Management Research Methodology: Integration of Principles, Methods and Techniques", 2016)

"Probability of not rejecting the null hypothesis when it's false." (Geoff Cumming, "Understanding The New Statistics", 2013)

Data Science: Supervised Learning (Definitions)

"A training paradigm where the neural network is presented with an input pattern and a desired output pattern. The desired output is compared with the neural network output, and the error information is used to adjust the connection weights." (Joseph P Bigus, "Data Mining with Neural Networks: Solving Business Problems from Application Development to Decision Support", 1996)

"Learning in which a system is trained by using a teacher to show the system the desired response to an input stimulus, usually in the form of a desired output." (Guido J Deboeck and Teuvo Kohonen, "Visual explorations in finance with self-organizing maps", 2000)

"learning with a teacher; learning scheme in which the average expected difference between wanted output for training samples, and the true output, respectively, is decreased." (Teuvo Kohonen, "Self-Organizing Maps" 3rd Ed., 2001)

"Supervised learning, or learning from examples, refers to systems that are trained instead of programmed with a set of examples, that is, a set of input-output pairs." (Tomaso Poggio & Steve Smale, "The Mathematics of Learning: Dealing with Data", Notices of the AMS, 2003)

"Methods, which use a response variable to guide the analysis." (Glenn J Myatt, "Making Sense of Data: A Practical Guide to Exploratory Data Analysis and Data Mining", 2006)

"A learning method in which there are two distinct phases to the operation. In the first phase each possible solution to a problem is assessed based on the input signal that is propagated through the system producing output respond. The actual respond produced is then compared with a desired response, generating error signals that are then used as a guide to solve the given problems using supervised learning algorithms". (Masoud Mohammadian, "Supervised Learning of Fuzzy Logic Systems", 2009)

"The set of learning algorithms in which the samples in the training dataset are all labelled." (Jun Jiang & Horace H S Ip, "Active Learning with SVM", Encyclopedia of Artificial Intelligence, 2009) 

"type of learning where the objective is to learn a function that associates a desired output (‘label’) to each input pattern. Supervised learning techniques require a training dataset of examples with their respective desired outputs. Supervised learning is traditionally divided into regression (the desired output is a continuous variable) and classification (the desired output is a class label)." (Óscar Pérez & Manuel Sánchez-Montañés, "Class Prediction in Test Sets with Shifted Distributions", 2009)

"Supervised learning is a type of machine learning that requires labeled training data." (Ivan Idris, "Python Data Analysis", 2014)

"Supervised learning refers to an approach that teaches the system to detect or match patterns in data based on examples it encounters during training with sample data." (Judith S Hurwitz, "Cognitive Computing and Big Data Analytics", 2015)

"The knowledge is obtained through a training which includes a data set called the training sample which is structured according to the knowledge base supported by human experts as physicians in medical context, and databases. It is assumed that the user knows beforehand the classes and the instances of each class." (Nuno Pombo et al, "Machine Learning Approaches to Automated Medical Decision Support Systems", 2015)

"In supervised learning, a machine learning program is trained with sample items or documents that are labeled by category, and the program learns to assign new items to the correct categories." (Robert J Glushko, "The Discipline of Organizing: Professional Edition" 4th Ed., 2016)

"A form of machine learning in which the goal is to learn a function that maps from a set of input attribute values for an instance to an estimate of the missing value for the target attribute of the same instance." (John D Kelleher & Brendan Tierney, "Data science", 2018)

"A supervised learning algorithm applies a known set of input data and drives a model to produce reasonable predictions for responses to new data. Supervised learning develops predictive models using classification and regression techniques." (Soraya Sedkaoui, "Big Data Analytics for Entrepreneurial Success", 2018)

"It consists in learning from data with a known-in-advance outcome that is predicted based on a set of inputs, referred to as 'features'." (Iva Mihaylova, "Applications of Artificial Neural Networks in Economics and Finance", 2018)

"Supervised learning is the data mining task of inferring a function from labeled training data." (Dharmendra S Rajput et al, "Investigation on Deep Learning Approach for Big Data: Applications and Challenges", 2018)

"A particular form of learning process that takes place under supervision and that affects the training of an artificial neural networks." (Gaetano B Ronsivalle & Arianna Boldi, "Artificial Intelligence Applied: Six Actual Projects in Big Organizations", 2019)

"A type of machine learning in which output datasets train the machine to generate the desired algorithms, like a teacher supervising a student." (Kirti R Bhatele et al, "The Role of Artificial Intelligence in Cyber Security", 2019)

"In this learning, the model needs a labeled data for training. The model knows in advance the answer to the questions it must predict and tries to learn the relationship between input and output." (Aman Kamboj et al, "EarLocalizer: A Deep-Learning-Based Ear Localization Model for Side Face Images in the Wild", 2019)

"A machine learning task designed to learn a function that maps an input onto an output based on a set of training examples (training data). Each training example is a pair consisting of a vector of inputs and an output value. A supervised learning algorithm analyzes the training data and infers a mapping function. A simple example of supervised learning is a regression model." (Timofei Bogomolov et al, "Identifying Patterns in Fresh Produce Purchases: The Application of Machine Learning Techniques", 2020)

"Supervised algorithms mean that a system is developed or modeled on predetermined set of sample data." (Neha Garg & Kamlesh Sharma, "Machine Learning in Text Analysis", 2020)

"A machine learning technique that involves providing a machine with data that is labeled." (Sujata Ramnarayan, "Marketing and Artificial Intelligence: Personalization at Scale", 2021)

"It is machine learning algorithm in which the model learns from ample amount of available labeled data to predict the class of unseen instances." (Gunjan Ansari et al, "Natural Language Processing in Online Reviews", 2021)

"Supervised learning aims at developing a function for a set of labeled data and outputs." (Hari K Kondaveeti et al, "Deep Learning Applications in Agriculture: The Role of Deep Learning in Smart Agriculture", 2021)

"The supervised learning algorithms are trained with a complete set of data and thus, the supervised learning algorithms are used to predict/forecast." (M Govindarajan, "Big Data Mining Algorithms", 2021)

"Supervised Learning is a type of machine learning in which an algorithm takes a labelled data set (data that’s been organized and described), deduces key features characterizing each label, and learns to recognize them in new unseen data." (Accenture)

15 May 2018

Data Science: Artificial Neural Network (Definitions)

"An artificial neural network (or simply a neural network) is a biologically inspired computational model that consists of processing elements (neurons) and connections between them, as well as of training and recall algorithms." (Nikola K Kasabov, "Foundations of Neural Networks, Fuzzy Systems, and Knowledge Engineering", 1996)

"Biologically inspired computational model consisting of processing elements (called neurons) and connections between them with coefficients (weights) bound to the connections, which constitute the neuronal structure. Training and recall algorithms are also attached to the structure." (Nikola K Kasabov, "Foundations of Neural Networks, Fuzzy Systems, and Knowledge Engineering", 1996)

"massively parallel interconnected network of simple (usually adaptive) elements and their hierarchical organizations, intended to interact with the objects of the real world in the same way as the biological nervous systems do. In a more general sense, artificial neural networks also encompass abstract schemata, such as mathematical estimators and systems of symbolic rules, constructed automatically from masses of examples, without heuristic design or other human intervention. Such schemata are supposed to describe the operation of biological or artificial neural networks in a highly idealized form and define certain performance limits." (Teuvo Kohonen, "Self-Organizing Maps" 3rd Ed., 2001)

"A collaboration of simple, primitive processing elements that self-organize and self-optimize to achieve computation goals. While these occur in biological systems, in this context we usually mean artificial neural networks such as might be used in optical character recognition applications." (Bruce P Douglass, "Real-Time Agility", 2009)

"An artificial neural network, often just called a “neural network” (NN), is an interconnected group of artificial neurons that uses a mathematical model or computational model for information processing based on a connectionist approach to computation. Knowledge is acquired by the network from its environment through a learning process, and interneuron connection strengths (synaptic weighs) are used to store the acquired knowledge." (Larbi Esmahi et al, "Adaptive Neuro-Fuzzy Systems", Encyclopedia of Artificial Intelligence, 2009)

"An interconnected group of units or neurons that uses a mathematical model for information processing based on a connectionist approach to computation." (Soledad Delgado et al, "Growing Self-Organizing Maps for Data Analysis", Encyclopedia of Artificial Intelligence, 2009)

"Artificial neural networks (ANNs) are non-linear mapping structures based on the function of the human brain. They are powerful tools for modeling, especially when the underlying data relationship is unknown." (Siddhartha Bhattacharjee et al, "Quantum Backpropagation Neural Network Approach for Modeling of Phenol Adsorption from Aqueous Solution by Orange Peel Ash", 2013)

"A computer representation of knowledge that attempts to mimic the neural networks of the human body" (Nell Dale & John Lewis, "Computer Science Illuminated" 6th Ed., 2015)

"a massively parallel distributed processor made up of simple processing units, which has a natural propensity for storing experimental knowledge and making it available for use." (Pablo Escandell-Montero et al, "Artificial Neural Networks in Physical Therapy", 2015)

"Computational models inspired by brain's nervous systems which are capable of machine learning and pattern recognition. ANN are composed by simple, and highly interconnected processing elements that process information by their dynamic state response to external inputs." (Nuno Pombo et al, "Machine Learning Approaches to Automated Medical Decision Support Systems", 2015)

"Computational models inspired by the properties of biological nervous systems. Usually composed of layers of highly interconnected simple processing units, they are characterised by learning capabilities and can be implemented in software and hardware." (D T Pham & M Castellani, "The Bees Algorithm as a Biologically Inspired Optimisation Method", 2015)

"Computer models of interconnected neurons that can be trained to carry out pattern recognition and other low-level cognitive functions through supervised or unsupervised of learning." (Eitan Gross, "Stochastic Neural Network Classifiers", Encyclopedia of Information Science and Technology 3rd Ed., 2015)

"Is non-parametric tool that learns from the surroundings, retains the learning and uses it subsequently." (Kandarpa K Sarma, "Learning Aided Digital Image Compression Technique for Medical Application", 2016)

"A computational graph for machine learning or simulation of a biological neural network (brain)." (Hobson Lane et al, "Natural Language Processing in Action: Understanding, analyzing, and generating text with Python", 2019)

"A machine learning algorithm that is created by mimicking the information transmission and problem-solving mechanism in the human brain." (Tolga Ensari et al, "Overview of Machine Learning Approaches for Wireless Communication", 2019)

"Information elaboration system, software, or hardware that is based on the biological nervous systems, and it is composed of code units called 'nodes' or 'artificial neurons'." (Gaetano B Ronsivalle & Arianna Boldi, "Artificial Intelligence Applied: Six Actual Projects in Big Organizations", 2019)

"A predictive computer algorithm inspired by the biology of the human brain that can learn linear and non-linear functions from data. Artificial neural networks are particularly useful when the complexity of the data or the modelling task makes the design of a function that maps inputs to outputs by hand impractical." (Timofei Bogomolov et al, "Identifying Patterns in Fresh Produce Purchases: The Application of Machine Learning Techniques", 2020)

"An artificial neural network is a collection of neurons connected by weights." (Alex Thomas, "Natural Language Processing with Spark NLP", 2020)

"Is a computational model based on the structure and functions of biological neural networks." (Heorhii Kuchuk et al, "Application of Deep Learning in the Processing of the Aerospace System's Multispectral Images", 2020)

"It mimics animal neural networks and useful in taking some action by observing some example instead of being explicitly programmed." (Shouvik Chakraborty & Kalyani Mali, "An Overview of Biomedical Image Analysis From the Deep Learning Perspective", 2020)

"An artificial neural network is based on a simplification of neurons in an animal brain which is a group of interconnected neurons." (Hari K Kondaveeti et al, "Deep Learning Applications in Agriculture: The Role of Deep Learning in Smart Agriculture", 2021)

"Artificial neural networks (ANNs) are a type of computing system that is inspired by biological neural networks present in the animal brain." (Hari K Kondaveeti et al, "Deep Learning Applications in Agriculture: The Role of Deep Learning in Smart Agriculture", 2021)

"Artificial neural networks (ANNs), also simply called neural networks (NNs), are bionic systems of neurons for vaguely computing and responding like human brains. ANNs show their power in the field of prediction and classification for a long time by a black-box system. ANNs enter a new era with the assistance of GPU for deep learning nowadays." (Yuh-Wen Chen, "Social Network Analysis: Self-Organizing Map and WINGS by Multiple-Criteria Decision Making", 2021)

"It is a computing model based on the structure of the human brain with many interconnected processing nodes that model input-output relationships. The model is organized in layers of nodes that interconnect to each other." (Mário P Véstias, "Convolutional Neural Network", 2021)

"It is an information processing model inspired by the form of the brain in which biological nervous systems, such as the brain, process information." (Mehmet A Cifci, "Optimizing WSNs for CPS Using Machine Learning Techniques", 2021)

"An artificial neuron network (ANN) is a computing system patterned after the operation of neurons in the human brain." (Databricks) [source]

14 May 2018

Data Science: Reinforcement Learning (Just the Quotes)

"A neural network training method based on presenting input vector x and looking at the output vector calculated by the network. If it is considered 'good', then a 'reward' is given to the network in the sense that the existing connection weights get increased, otherwise the network is "punished"; the connection weights, being considered as 'not appropriately set,' decrease." (Nikola K Kasabov, "Foundations of Neural Networks, Fuzzy Systems, and Knowledge Engineering", 1996)

"A training paradigm where the neural network is presented with a sequence of input data, followed by a reinforcement signal." (Joseph P Bigus, "Data Mining with Neural Networks: Solving Business Problems from Application Development to Decision Support", 1996)

"learning mode in which adaptive changes of the parameters due to reward or punishment depend on the final outcome of a whole sequence of behavior. The results of learning are evaluated by some performance index." (Teuvo Kohonen, "Self-Organizing Maps" 3rd Ed., 2001)

"A learning method which interprets feedback from an environment to learn optimal sets of condition/response relationships for problem solving within that environment" (Pi-Sheng Deng, "Genetic Algorithm Applications to Optimization Modeling", Encyclopedia of Artificial Intelligence, 2009)

"A sub-area of machine learning concerned with how an agent ought to take actions in an environment so as to maximize some notion of long-term reward. Reinforcement learning algorithms attempt to find a policy that maps states of the world to the actions the agent ought to take in those states. Differently from supervised learning, in this case there is no target value for each input pattern, only a reward based of how good or bad was the action taken by the agent in the existent environment." (Marley Vellasco et al, "Hierarchical Neuro-Fuzzy Systems" Part II, Encyclopedia of Artificial Intelligence, 2009)

"a type of machine learning in which an agent learns, through its own experience, to navigate through an environment, choosing actions in order to maximize the sum of rewards." (Lisa Torrey & Jude Shavlik, "Transfer Learning",  2010)

"a machine learning technique whereby actions are associated with credits or penalties, sometimes with delay, and whereby, after a series of learning episodes, the learning agent has developed a model of which action to choose in a particular environment, based on the expectation of accumulated rewards." (Apostolos Georgas, "Scientific Workflows for Game Analytics", Encyclopedia of Business Analytics and Optimization", 2014)

"A type of machine learning in which the machine learns what to do by discovering through trial and error the way to maximize a reward." (Gloria Phillips-Wren, "Intelligent Systems to Support Human Decision Making", 2014)

"it stands, in the context of computational learning, for a family of algorithms aimed at approximating the best policy to play in a certain environment (without building an explicit model of it) by increasing the probability of playing actions that improve the rewards received by the agent." (Fernando S Oliveira, "Reinforcement Learning for Business Modeling", 2014)

"a special case of supervised learning in which the cognitive computing system receives feedback on its performance to guide it to a goal or good outcome." (Judith S Hurwitz, "Cognitive Computing and Big Data Analytics", 2015)

"The knowledge is obtained using rewards and punishments which there is an agent (learner) that acts autonomously and receives a scalar reward signal that is used to evaluate the consequences of its actions." (Nuno Pombo et al, "Machine Learning Approaches to Automated Medical Decision Support Systems", 2015)

"It is also known as learning with a critic. The agent takes a sequence of actions and receives a reward/penalty only at the very end, with no feedback during the intermediate actions. Using this limited information, the agent should learn to generate the actions to maximize the reward in later trials. For example, in chess, we do a set of moves, and at the very end, we win or lose the game; so we need to figure out what the actions that led us to this result were and correspondingly credit them." (Ethem Alpaydın, "Machine learning : the new AI", 2016)

"A learning algorithm for a robot or a software agent to take actions in an environment so as to maximize the sum of rewards through trial and error." (Tomohiro Yamaguchi et al, "Analyzing the Goal-Finding Process of Human Learning With the Reflection Subtask", 2018)

"Training/learning method aiming to automatically determine the ideal behavior within a specific context based on rewarding desired behaviors and/or punishing undesired one." (Ioan-Sorin Comşa et al, "Guaranteeing User Rates With Reinforcement Learning in 5G Radio Access Networks", 2019)

"Brach of the Artificial Intelligence field devoted to obtaining optimal control sequences for agents only by interacting with a concrete dynamical system." (Juan Parras & Santiago Zazo, "The Threat of Intelligent Attackers Using Deep Learning: The Backoff Attack Case", 2020)

"Machine learning approaches often used in robotics. A reward is used to teach a system a desired behavior." (Jörg Frochte et al, "Concerning the Integration of Machine Learning Content in Mechatronics Curricula", 2020)

"This area of deep learning includes methods which iterates over various steps in a process to get the desired results. Steps that yield desirable outcomes are content and steps that yield undesired outcomes are reprimanded until the algorithm is able to learn the given optimal process. In unassuming terms, learning is finished on its own or effort on feedback or content-based learning." (Amit K Tyagi & Poonam Chahal, "Artificial Intelligence and Machine Learning Algorithms", 2020)

"A machine learning paradigm that utilizes evaluative feedback to cultivate desired behavior." (Marten H L Kaas, "Raising Ethical Machines: Bottom-Up Methods to Implementing Machine Ethics", 2021)

"Is an area of machine learning that learn for the experience in order to maximize the rewards." (Walaa Alnasser et al, "An Overview on Protecting User Private-Attribute Information on Social Networks", 2021)

"Reinforcement learning is also a subset of AI algorithms which creates independent, self-learning systems through trial and error. Any positive action is assigned a reward and any negative action would result in a punishment. Reinforcement learning can be used in training autonomous vehicles where the goal would be obtaining the maximum rewards." (Vijayaraghavan Varadharajan & Akanksha Rajendra Singh, "Building Intelligent Cities: Concepts, Principles, and Technologies", 2021)

"Reinforcement Learning uses a kind of algorithm that works by trial and error, where the learning is enabled using a feedback loop of 'rewards' and 'punishments'. When the algorithm is fed a dataset, it treats the environment like a game, and is told whether it has won or lost each time it performs an action. In this way, reinforcement learning algorithms build up a picture of the 'moves' that result in success, and those that don't." (Accenture)

13 May 2018

Data Science: Self-Organizing Map (Definitions)

"A clustering neural net, with topological structure among cluster units." (Laurene V Fausett, "Fundamentals of Neural Networks: Architectures, Algorithms, and Applications", 1994)

"A self organizing map is a form of Kohonen network that arranges its clusters in a (usually) two-dimensional grid so that the codebook vectors (the cluster centers) that are close to each other on the grid are also close in the k-dimensional feature space. The converse is not necessarily true, as codebook vectors that are close in feature-space might not be close on the grid. The map is similar in concept to the maps produced by descriptive techniques such as multi-dimensional scaling (MDS)." (William J Raynor Jr., "The International Dictionary of Artificial Intelligence", 1999)

"result of a nonparametric regression process that is mainly used to represent high-dimensional, nonlinearly related data items in an illustrative, often two-dimensional display, and to perform unsupervised classification and clustering." (Teuvo Kohonen, "Self-Organizing Maps" 3rd Ed., 2001)

"a method of organizing and displaying textual information according to the frequency of occurrence of text and the relationship of text from one document to another." (William H Inmon, "Building the Data Warehouse", 2005)

"A type of unsupervised neural network used to group similar cases in a sample. SOMs are unsupervised (see supervised network) in that they do not require a known dependent variable. They are typically used for exploratory analysis and to reduce dimensionality as an aid to interpretation of complex data. SOMs are similar in purpose to Ic-means clustering and factor analysis." (David Scarborough & Mark J Somers, "Neural Networks in Organizational Research: Applying Pattern Recognition to the Analysis of Organizational Behavior", 2006)

"A method to learn to cluster input vectors according to how they are naturally grouped in the input space. In its simplest form, the map consists of a regular grid of units and the units learn to represent statistical data described by model vectors. Each map unit contains a vector used to represent the data. During the training process, the model vectors are changed gradually and then the map forms an ordered non-linear regression of the model vectors into the data space." (Atiq Islam et al, "CNS Tumor Prediction Using Gene Expression Data Part II", Encyclopedia of Artificial Intelligence, 2009)

"A neural-network method that reduces the dimensions of data while preserving the topological properties of the input data. SOM is suitable for visualizing high-dimensional data such as microarray data." (Emmanuel Udoh & Salim Bhuiyan, "C-MICRA: A Tool for Clustering Microarray Data", 2009)

"A neural network unsupervised method of vector quantization widely used in classification. Self-Organizing Maps are a much appreciated for their topology preservation property and their associated data representation system. These two additive properties come from a pre-defined organization of the network that is at the same time a support for the topology learning and its representation. (Patrick Rousset & Jean-Francois Giret, "A Longitudinal Analysis of Labour Market Data with SOM" Encyclopedia of Artificial Intelligence, 2009)

"A simulated neural network based on a grid of artificial neurons by means of prototype vectors. In an unsupervised training the prototype vectors are adapted to match input vectors in a training set. After completing this training the SOM provides a generalized K-means clustering as well as topological order of neurons." (Laurence Mukankusi et al, "Relationships between Wireless Technology Investment and Organizational Performance", 2009)

"A subtype of artificial neural network. It is trained using unsupervised learning to produce low dimensional representation of the training samples while preserving the topological properties of the input space." (Soledad Delgado et al, "Growing Self-Organizing Maps for Data Analysis", 2009)

"An unsupervised neural network providing a topology-preserving mapping from a high-dimensional input space onto a two-dimensional output space." (Thomas Lidy & Andreas Rauber, "Music Information Retrieval", 2009)

"Category of algorithms based on artificial neural networks that searches, by means of self-organization, to create a map of characteristics that represents the involved samples in a determined problem." (Paulo E Ambrósio, "Artificial Intelligence in Computer-Aided Diagnosis", 2009)

"Self-organizing maps (SOMs) are a data visualization technique which reduce the dimensions of data through the use of self-organizing neural networks." (Lluís Formiga & Francesc Alías, "GTM User Modeling for aIGA Weight Tuning in TTS Synthesis", Encyclopedia of Artificial Intelligence, 2009)

"SOFM [self-organizing feature map] is a data mining method used for unsupervised learning. The architecture consists of an input layer and an output layer. By adjusting the weights of the connections between input and output layer nodes, this method identifies clusters in the data." (Indranil Bose, "Data Mining in Tourism", 2009)

"The self-organizing map is a subtype of artificial neural networks. It is trained using unsupervised learning to produce low dimensional representation of the training samples while preserving the topological properties of the input space. The self-organizing map is a single layer feed-forward network where the output syntaxes are arranged in low dimensional (usually 2D or 3D) grid. Each input is connected to all output neurons. Attached to every neuron there is a weight vector with the same dimensionality as the input vectors. The number of input dimensions is usually a lot higher than the output grid dimension. SOMs are mainly used for dimensionality reduction rather than expansion." (Larbi Esmahi et al, "Adaptive Neuro-Fuzzy Systems", Encyclopedia of Artificial Intelligence, 2009)

"A type of neural network that uses unsupervised learning to produce two-dimensional representations of an input space." (DAMA International, "The DAMA Dictionary of Data Management", 2011)

"The Self-organizing map is a non-parametric and non-linear neural network that explores data using unsupervised learning. The SOM can produce output that maps multidimensional data onto a two-dimensional topological map. Moreover, since the SOM requires little a priori knowledge of the data, it is an extremely useful tool for exploratory analyses. Thus, the SOM is an ideal visualization tool for analyzing complex time-series data." (Peter Sarlin, "Visualizing Indicators of Debt Crises in a Lower Dimension: A Self-Organizing Maps Approach", 2012)

"SOMs or Kohonen networks have a grid topology, with unequal grid weights. The topology of the grid provides a low dimensional visualization of the data distribution." (Siddhartha Bhattacharjee et al, "Quantum Backpropagation Neural Network Approach for Modeling of Phenol Adsorption from Aqueous Solution by Orange Peel Ash", 2013)

"An unsupervised neural network widely used in exploratory data analysis and to visualize multivariate object relationships." (Manuel Martín-Merino, "Semi-Supervised Dimension Reduction Techniques to Discover Term Relationships", 2015)

"ANN used for visualizing low-dimensional views of high-dimensional data." (Pablo Escandell-Montero et al, "Artificial Neural Networks in Physical Therapy", 2015)

"Is a unsupervised learning ANN, which means that no human intervention is needed during the learning and that little needs to be known about the characteristics of the input data." (Nuno Pombo et al, "Machine Learning Approaches to Automated Medical Decision Support Systems", 2015)

"A kind of artificial neural network which attempts to mimic brain functions to provide learning and pattern recognition techniques. SOM have the ability to extract patterns from large datasets without explicitly understanding the underlying relationships. They transform nonlinear relations among high dimensional data into simple geometric connections among their image points on a low-dimensional display." (Felix Lopez-Iturriaga & Iván Pastor-Sanz, "Using Self Organizing Maps for Banking Oversight: The Case of Spanish Savings Banks", 2016)

"Neural network which simulated some cerebral functions in elaborating visual information. It is usually used to classify a large amount of data." (Gaetano B Ronsivalle & Arianna Boldi, "Artificial Intelligence Applied: Six Actual Projects in Big Organizations", 2019)

"Classification technique based on unsupervised-learning artificial neural networks allowing to group data into clusters." Julián Sierra-Pérez & Joham Alvarez-Montoya, "Strain Field Pattern Recognition for Structural Health Monitoring Applications", 2020)

"It is a type of artificial neural network (ANN) trained using unsupervised learning for dimensionality reduction by discretized representation of the input space of the training samples called as map." (Dinesh Bhatia et al, "A Novel Artificial Intelligence Technique for Analysis of Real-Time Electro-Cardiogram Signal for the Prediction of Early Cardiac Ailment Onset", 2020)

"Being a particular type of ANNs, the Self Organizing Map is a simple mapping from inputs: attributes directly to outputs: clusters by the algorithm of unsupervised learning. SOM is a clustering and visualization technique in exploratory data analysis." (Yuh-Wen Chen, "Social Network Analysis: Self-Organizing Map and WINGS by Multiple-Criteria Decision Making", 2021)

12 May 2018

Data Science: Backpropagation (Definitions)

"A learning algorithm for multilayer neural nets based on minimizing the mean, or total, squared error." (Laurene V Fausett, "Fundamentals of Neural Networks: Architectures, Algorithms, and Applications", 1994)

"A learning scheme by which a multi-layer feedforward network is organized for pattern recognition or classification utilizing an external teacher, and error feedback (or propagation)." (Guido J Deboeck and Teuvo Kohonen, "Visual explorations in finance with self-organizing maps", 2000)

"weight-vector optimization method used in multilayered feed-forward networks. The corrective steps arc made starting at the output layer and proceeding toward the input layer." (Teuvo Kohonen, "Self-Organizing Maps" 3rd Ed., 2001)

"A class of feed-forward neural networks used for classification, forecasting, and estimation. Backpropagation is the process by which connection weights between neurons are modified using a backward pass of error derivatives." (David Scarborough & Mark J Somers, "Neural Networks in Organizational Research: Applying Pattern Recognition to the Analysis of Organizational Behavior", 2006)

"A method for training a neural network by adjusting the weights using errors between the current prediction and the training set." (Glenn J Myatt, "Making Sense of Data: A Practical Guide to Exploratory Data Analysis and Data Mining", 2006)

"A supervised learning algorithm used to train artificial neural networks, where the network learns from many inputs, similar to the way a child learns to identify a bird from examples of birds and birds attributes." (Eitan Gross, "Stochastic Neural Network Classifiers", 2015)

"A learning algorithm for artificial neural networks used for supervised learning, where connection weights are iteratively updated to decrease the approximation error at the output units." (Ethem Alpaydın, "Machine learning: the new AI", 2016)

"Learning algorithm that optimizes a neural network by gradient descent to minimize a cost function and improve performance." (Terrence J Sejnowski, "The Deep Learning Revolution", 2018)

"The backpropagation algorithm is an ML algorithm used to train neural networks. The algorithm calculates for each neuron in a network the contribution |  the neuron makes to the error of the network. Using this error calculation for each neuron it is possible to update the weights on the inputs to each neuron so as to reduce the overall error of the network. The backpropagation algorithm is so named because it works in a two stage process. In the first stage an instance is input to the network and the information flows forward through the network until the network generates a prediction for that instance. In the second stage the error of the network on that instance is calculated by comparing the network's prediction to the correct output for that instance (as specified by the training data) and then this error is then shared back (or backpropagated) through the neurons in the network on a layer by layer basis beginning at the output layer." (John D Kelleher & Brendan Tierney, "Data Science", 2018)

"Backpropagation is short for 'backward propagation of errors'. Backpropagation in convolutional neural networks is a way of training these networks based on a known, desired output for a specific sample case." (Accenture)

11 May 2018

Application Support: One Database, Two Vendors, No Love Story or Maybe…

Data Warehousing

Introduction

Situation: An organization has several BI tools provisioned with data from the same data warehouse (DW), the BI infrastructure being supported by the same service provider (vendor). The organization wants to adopt a new BI technology, though for it must be brought another vendor into the picture. The data the tool requires are already available in the DW, though the DW needs to be extended with logic and other components to support the new tool. This means that two vendors will be active in the same DW, more generally in the same environment.

Question(s): What is the best approach for making this work? Which are the challenges for making it work, considering that two vendors?


Preliminary

    When you ask IT people about this situation, many will tell you that’s not a good idea, being circumspect at having two vendors within the same environment. Some will recall previous experience in which things went really bad, escalated to some degree. They will even show their disagreement through body language or increase tonality. Even if they had also good experiences with having two vendors support the same environment, the negative experiences will prevail. It’s the typical reaction to an idea when something that caused considerable trouble is recalled. This behavior is understandable as generally human tend to remember more the issues they had, rather than successes. Problems leave deeper marks than success, especially when challenges are seen as burdens.

    Reacting defensively is a result of the “I’ve been burned once” syndrome. People react adversely and tend to avoid situations in which they were burned, instead of dealing with them, instead of recognizing which were the circumstances that lead to the situation in the first place, of recognizing opportunities for healing and raising above the challenges.


   Personally, at a first glance, the caution would make me advise as well against having two or more vendors playing within same playground. I had my plate of extreme cases in which something went wrong and the vendors started acting like kids. Parents (in general people who work with children) know what I’m talking about, children don’t like to share their toys and parents often find themselves in the position of mediating between them. When the toy get’s broken it’s easy to blame other kid for it, same as somebody else must put the toy at its place, because that somebody played the last time with it. It’s a mix between I’m in charge and the blame game. Who needs that?

  At second sight, if parents made it, why wouldn’t professionals succeed in making two vendors work together? Sure, parents have more practice in dealing with kids, have such situations on a daily basis, and there are fewer variables to think about it… I have seen vendors sitting together until they come up with a solution, I’ve seen vendors open to communicate, putting the customer on the first place, even if that meant living the ego behind. Where there’s a will there’s a way.


The Solution Space

    In IT there are seldom general recipes that always lead to success, and whether a solution works or not depends on a serious of factors – environment, skills, communication, human behavior and quite often chance, the chance of doing the right thing at the right time. However, the recipe can be used as a starting point, eventually to define the best scenario, what will happen when everything goes well. At the opposite side there is the worst scenario, what will happen when everything goes south. These two opposite scenarios are in general the frame in which a solution can be defined.

    Within this frame one can add several other reference points or paths, and these are made of the experience of people handling and experiencing similar situations – what worked, what didn’t, what could work, what are the challenges, and so on. In general, people’s experience and knowledge prove to be a good estimator in decision making, and even if relative, it proves some insight into the problem at hand.


    Let’s reconsider the parents put in the situation of dealing with children fighting for the same toy, though from the perspective of all the toys available to play with. There are several options available: the kids could take (supervised) turn in playing with the toys, fact that could be a win-win situation if they are willing to cooperate. One can take the toys (temporarily) away, though this could lead to other issues. One can reaffirm who’s the owner of each toy, each kid being allowed to play only with his toy. One could buy a second toy, and thus brake the bank even if this will not make the issue entirely go away. Probably there are other solutions inventive parents might find.

    Similarly, in the above stated problem, one option, and maybe the best, is having the vendors share ownership for the DW by finding a way to work together. Defining the ownership for each tool can alleviate some of the problems but not all, same as building a second DW. We can probably all agree that taking the tools away is not the right thing to do, and even if it’s a solution, it doesn’t support the purpose.


Sharing Ownership

    Complex IT environments like the one of a DW depend on vendors’ capability of working together in reaching the same goal, even if in play are different interests. This presumes the disposition of the parties in relinquishing some control, sharing responsibilities. Unfortunately, not all vendors are willing to do that. That’s the point where imaginary obstacles are built, is where effort needs to be put to eliminate such obstacles.

    When working together, often one of the parties must play the coordinator role. In theory, this role can be played by any of the vendors, and the roles can even change after case. Another approach is when the coordinator role can be taken also by a person or a board from the customer side. In case of a data warehouse it can be an IT professional, a Project Manager or a BI Competency Center (BICC) . This would allow to smoothly coordinate the activities, as well to mediate the communication and other types of challenges faced.


    How will ownership sharing work? Let’s suppose vendor A wants to change something in the infrastructure. The change is first formulated, shortly reviewed, and approved by both vendors and customer, and will then be implemented and documented by vendor A as needed. Vendor B is involved in the process by validating the concept and reviewing the documentation, its involvement being minimized. There can be still some delays in the process, though the overhead is somehow minimized. There will be also scenarios in which vendor B needs only to be informed that a change has occurred, or sometimes is enough if a change was properly documented.

    This approach involves also a greater need for documentation, versioning, established processes, their role being to facilitate the communication and track the changes occurred in the environment.


Splitting Ownership

    Splitting the ownership involves setting clear boundaries and responsibilities within which each vendor can perform the work. One is forced thus to draw a line and say which components or activities belong to each vendor. 

    The architecture of existing solutions makes it sometimes hard to split the ownership when the architecture was not designed for it. A solution would be to redesign the whole architecture, though even then might not be possible to draw a clear line when there are grey areas. One needs eventually to consider the advantages and disadvantages and decide to which vendor the responsibility suits best.


    For example, in the context of a DW security can be enforced via schemas within same or different databases, though there are also objects (e.g. tables with basis data) used by multiple applications. One of the vendors (vendor A) will get the ownership of the objects, thus when vendor B needs a change to those table, it must require the change to vendor A. Once the changes are done the vendor B needs to validate the changes, and if there are problems further communication occurs. Per total this approach will take more time than if the vendor B would have done alone the changes. However, it works even if it comes with some challenges.

    There’s also the possibility to give vendor B temporary permissions to do the changes, fact that will shorten the effort needed. The vendor A will still be in charge, and will have to prove the documentation, and do eventually some validation as well.


Separating Ownership

    Giving each vendor its own playground is a costly solution, though it can be the only solution in certain scenarios. For example, when an architecture is supposed to replace (in time) another, or when the existing architecture has certain limitations. In the context of a DW this involves duplicating the data loads, the data themselves, as well logic, eventually processes, and so on.

    Pushing this just to solve a communication problem is the wrong thing to do. What happens if a third or a fourth vendor joins the party? Would it be for each vendor a new environment created? Hopefully, not…

    On the other side, there are also vendors that don’t want to relinquish the ownership, and will play their cards not to do it. The overhead of dealing with such issues may surpass in extremis the costs of having a second environment. In the end the final decision has the customer.


Hybrid Approach


    A hybrid between sharing and splitting ownership can prove to give the best from the two scenarios. It’s useful and even recommended to define the boundaries of work for each vendor, following to share ownership on the areas where there’s an intersection of concerns, the grey areas. For sensitive areas there could be some restrictions in cooperation.

    A hybrid solution can involve as well splitting some parts of the architecture, though the performance and security are mainly the driving factors.


Conclusion

   I wanted with this post to make the reader question some of the hot-brained decisions made when two or more vendors are involved in the supporting an architecture. Even if the problem is put within the context of a DW it’s occurrence extends far beyond this context. We are enablers and problem solvers. Instead of avoiding challenges we should better make sure that we’re removing or minimizing the risks. 

Related Posts Plugin for WordPress, Blogger...