book

Artificial Intelligence By Example - Second Edition

Name: Artificial Intelligence By Example - Second Edition
Author: Denis Rothman
ISBN: 9781839211539

by Denis Rothman

February 2020

Beginner

578 pages

13h 16m

English

Packt Publishing

Read now

Unlock full access

Preface
Who this book is forWhat this book coversTo get the most out of this bookGet in touch
Getting Started with Next-Generation Artificial Intelligence through Reinforcement Learning
Reinforcement learning conceptsHow to adapt to machine thinking and become an adaptive thinkerOvercoming real-life issues using the three-step approachStep 1 – describing a problem to solve: MDP in natural languageWatching the MDP agent at workStep 2 – building a mathematical model: the mathematical representation of the Bellman equation and MDPFrom MDP to the Bellman equationStep 3 – writing source code: implementing the solution in PythonThe lessons of reinforcement learningHow to use the outputsPossible use casesMachine learning versus traditional applicationsSummaryQuestionsFurther reading
Building a Reward Matrix – Designing Your Datasets
Designing datasets – where the dream stops and the hard work beginsDesigning datasetsUsing the McCulloch-Pitts neuronThe McCulloch-Pitts neuronThe Python-TensorFlow architectureLogistic activation functions and classifiersOverall architectureLogistic classifierLogistic functionSoftmaxSummaryQuestionsFurther reading
Machine Intelligence – Evaluation Functions and Numerical Convergence
Tracking down what to measure and deciding how to measure itConvergenceImplicit convergenceNumerically controlled gradient descent convergenceEvaluating beyond human analytic capacityUsing supervised learning to evaluate a result that surpasses human analytic capacitySummaryQuestionsFurther reading
Optimizing Your Solutions with K-Means Clustering
Dataset optimization and controlDesigning a dataset and choosing an ML/DL modelApproval of the design matrixImplementing a k-means clustering solutionThe visionThe dataThe strategyThe k-means clustering programThe mathematical definition of k-means clusteringThe Python programSaving and loading the modelAnalyzing the resultsBot virtual clusters as a solutionThe limits of the implementation of the k-means clustering algorithmSummaryQuestionsFurther reading
How to Use Decision Trees to Enhance K-Means Clustering
Unsupervised learning with KMC with large datasetsIdentifying the difficulty of the problemNP-hard – the meaning of PNP-hard – the meaning of non-deterministicImplementing random sampling with mini-batchesUsing the LLNThe CLTUsing a Monte Carlo estimatorTrying to train the full training datasetTraining a random sample of the training datasetShuffling as another way to perform random samplingChaining supervised learning to verify unsupervised learningPreprocessing raw dataA pipeline of scripts and ML algorithmsStep 1 – training and exporting data from an unsupervised ML algorithmStep 2 – training a decision treeStep 3 – a continuous cycle of KMC chained to a decision treeRandom forests as an alternative to decision treesSummaryQuestionsFurther reading
Innovating AI with Google Translate
Understanding innovation and disruption in AIIs AI disruptive?AI is based on mathematical theories that are not newNeural networks are not newLooking at disruption – the factors that are making AI disruptiveCloud server power, data volumes, and web sharing of the early 21st centuryPublic awarenessInventions versus innovationsRevolutionary versus disruptive solutionsWhere to start?Discover a world of opportunities with Google TranslateGetting startedThe programThe headerImplementing Google's translation serviceGoogle Translate from a linguist's perspectivePlaying with the toolLinguistic assessment of Google TranslateAI as a new frontierLexical field and polysemyExploring the frontier – customizing Google Translate with a Python programk-nearest neighbor algorithmImplementing the KNN algorithmThe knn_polysemy.py programImplementing the KNN function in Google_Translate_Customized.pyConclusions on the Google Translate customized experimentThe disruptive revolutionary loopSummaryQuestionsFurther reading
Optimizing Blockchains with Naive Bayes
Part I – the background to blockchain technologyMining bitcoinsUsing cryptocurrencyPART II – using blockchains to share information in a supply chainUsing blockchains in the supply chain networkCreating a blockExploring the blocksPart III – optimizing a supply chain with naive Bayes in a blockchain processA naive Bayes exampleThe blockchain anticipation noveltyThe goal – optimizing storage levels using blockchain dataImplementation of naive Bayes in PythonGaussian naive BayesSummaryQuestionsFurther reading
Solving the XOR Problem with a Feedforward Neural Network
The original perceptron could not solve the XOR functionXOR and linearly separable modelsLinearly separable modelsThe XOR limit of a linear model, such as the original perceptronBuilding an FNN from scratchStep 1 – defining an FNNStep 2 – an example of how two children can solve the XOR problem every dayImplementing a vintage XOR solution in Python with an FNN and backpropagationA simplified version of a cost function and gradient descentLinear separability was achievedApplying the FNN XOR function to optimizing subsets of dataSummaryQuestionsFurther reading
Abstract Image Classification with Convolutional Neural Networks (CNNs)
Introducing CNNsDefining a CNNInitializing the CNNAdding a 2D convolution layerKernelShapeReLUPoolingNext convolution and pooling layerFlatteningDense layersDense activation functionsTraining a CNN modelThe goalCompiling the modelThe loss functionThe Adam optimizerMetricsThe training datasetData augmentationLoading the dataThe testing datasetData augmentation on the testing datasetLoading the dataTraining with the classifierSaving the modelNext stepsSummaryQuestionsFurther reading and references

Conceptual Representation Learning
Generating profit with transfer learningThe motivation behind transfer learningInductive thinkingInductive abstractionThe problem AI needs to solveThe Γ gap conceptLoading the trained TensorFlow 2.x modelLoading and displaying the modelLoading the model to use itDefining a strategyMaking the model profitable by using it for another problemDomain learningHow to use the programsThe trained models used in this sectionThe trained model programGap – loaded or underloadedGap – jammed or open lanesGap datasets and subsetsGeneralizing the Γ (the gap conceptual dataset)The motivation of conceptual representation learning metamodels applied to dimensionalityThe curse of dimensionalityThe blessing of dimensionalitySummaryQuestionsFurther reading
Combining Reinforcement Learning and Deep Learning
Planning and scheduling today and tomorrowA real-time manufacturing processAmazon must expand its services to face competitionA real-time manufacturing revolutionCRLMM applied to an automated apparel manufacturing processAn apparel manufacturing processTraining the CRLMMGeneralizing the unit training datasetFood conveyor belt processing – positive pγ and negative nγ gapsRunning a prediction programBuilding the RL-DL-CRLMMA circular processImplementing a CNN-CRLMM to detect gaps and optimizeQ-learning – MDPMDP inputs and outputsThe optimizerThe optimizer as a regulatorFinding the main target for the MDP functionA circular model – a stream-like system that never starts nor endsSummaryQuestionsFurther reading
AI and the Internet of Things (IoT)
The public service projectSetting up the RL-DL-CRLMM modelApplying the model of the CRLMMThe datasetUsing the trained modelAdding an SVM functionMotivation – using an SVM to increase safety levelsDefinition of a support vector machinePython functionRunning the CRLMMFinding a parking spaceDeciding how to get to the parking lotSupport vector machineThe itinerary graphThe weight vectorSummaryQuestionsFurther reading
Visualizing Networks with TensorFlow 2.x and TensorBoard
Exploring the output of the layers of a CNN in two steps with TensorFlowBuilding the layers of a CNNProcessing the visual output of the layers of a CNNAnalyzing the visual output of the layers of a CNNAnalyzing the accuracy of a CNN using TensorBoardGetting started with Google ColaboratoryDefining and training the modelIntroducing some of the measurementsSummaryQuestionsFurther reading
Preparing the Input of Chatbots with Restricted Boltzmann Machines (RBMs) and Principal Component Analysis (PCA)
Defining basic terms and goalsIntroducing and building an RBMThe architecture of an RBMAn energy-based modelBuilding the RBM in PythonCreating a class and the structure of the RBMCreating a training function in the RBM classComputing the hidden units in the training functionRandom sampling of the hidden units for the reconstruction and contractive divergenceReconstructionContrastive divergenceError and energy functionRunning the epochs and analyzing the resultsUsing the weights of an RBM as feature vectors for PCAUnderstanding PCAMathematical explanationUsing TensorFlow's Embedding Projector to represent PCAAnalyzing the PCA to obtain input entry points for a chatbotSummaryQuestionsFurther reading
Setting Up a Cognitive NLP UI/CUI Chatbot
Basic conceptsDefining NLUWhy do we call chatbots "agents"?Creating an agent to understand DialogflowEntitiesIntentsContextAdding fulfillment functionality to an agentDefining fulfillmentEnhancing the cogfilmdr agent with a fulfillment webhookGetting the bot to work on your websiteMachine learning agentsUsing machine learning in a chatbotSpeech-to-textText-to-speechSpellingWhy are these machine learning algorithms important?SummaryQuestionsFurther reading
Improving the Emotional Intelligence Deficiencies of Chatbots
From reacting to emotions, to creating emotionsSolving the problems of emotional polysemyThe greetings problem exampleThe affirmation exampleThe speech recognition fallacyThe facial analysis fallacySmall talkCourtesyEmotionsData loggingCreating emotionsRNN research for future automatic dialog generationRNNs at workRNN, LSTM, and vanishing gradientsText generation with an RNNVectorizing the textBuilding the modelGenerating textSummaryQuestionsFurther reading
Genetic Algorithms in Hybrid Neural Networks
Understanding evolutionary algorithmsHeredity in humansOur cellsHow heredity worksEvolutionary algorithmsGoing from a biological model to an algorithmBasic conceptsBuilding a genetic algorithm in PythonImporting the librariesCalling the algorithmThe main functionThe parent generation processGenerating a parentFitnessDisplay parentCrossover and mutationProducing generations of childrenSummary codeUnspecified target to optimize the architecture of a neural network with a genetic algorithmA physical neural networkWhat is the nature of this mysterious S-FNN?Calling the algorithm cellFitness cellga_main() cellArtificial hybrid neural networksBuilding the LSTMThe goal of the modelSummaryQuestionsFurther reading
Neuromorphic Computing
Neuromorphic computingGetting started with NengoInstalling Nengo and Nengo GUICreating a Python programA Nengo ensembleNengo neuron typesNengo neuron dimensionsA Nengo nodeConnecting Nengo objectsVisualizing dataProbesApplying Nengo's unique approach to critical AI research areasSummaryQuestionsReferencesFurther reading
Quantum Computing
The rising power of quantum computersQuantum computer speedDefining a qubitRepresenting a qubitThe position of a qubitRadians, degrees, and rotationsThe Bloch sphereComposing a quantum scoreQuantum gates with QuirkA quantum computer score with QuirkA quantum computer score with IBM QA thinking quantum computerRepresenting our mind's conceptsExpanding MindX's conceptual representationsThe MindX experimentPreparing the dataTransformation functions – the situation functionTransformation functions – the quantum functionCreating and running the scoreUsing the outputSummaryQuestionsFurther reading
Answers to the Questions
Chapter 1 – Getting Started with Next-Generation Artificial Intelligence through Reinforcement LearningChapter 2 – Building a Reward Matrix – Designing Your DatasetsChapter 3 – Machine Intelligence – Evaluation Functions and Numerical ConvergenceChapter 4 – Optimizing Your Solutions with K-Means ClusteringChapter 5 – How to Use Decision Trees to Enhance K-Means ClusteringChapter 6 – Innovating AI with Google TranslateChapter 7 – Optimizing Blockchains with Naive BayesChapter 8 – Solving the XOR Problem with a Feedforward Neural NetworkChapter 9 – Abstract Image Classification with Convolutional Neural Networks (CNNs)Chapter 10 – Conceptual Representation LearningChapter 11 – Combining Reinforcement Learning and Deep LearningChapter 12 – AI and the Internet of ThingsChapter 13 – Visualizing Networks with TensorFlow 2.x and TensorBoardChapter 14 – Preparing the Input of Chatbots with Restricted Boltzmann Machines (RBMs) and Principal Component Analysis (PCA)Chapter 15 – Setting Up a Cognitive NLP UI/CUI ChatbotChapter 16 – Improving the Emotional Intelligence Deficiencies of ChatbotsChapter 17 – Genetic Algorithms in Hybrid Neural NetworksChapter 18 – Neuromorphic ComputingChapter 19 – Quantum Computing
Other Books You May Enjoy
Index

Content preview from Artificial Intelligence By Example - Second Edition

2 Building a Reward Matrix – Designing Your Datasets

Experimenting and implementation comprise the two main approaches of artificial intelligence. Experimenting largely entails trying ready-to-use datasets and black box, ready-to-use Python examples. Implementation involves preparing a dataset, developing preprocessing algorithms, and then choosing a model, the proper parameters, and hyperparameters.

Implementation usually involves white box work that entails knowing exactly how an algorithm works and even being able to modify it.

In Chapter 1, Getting Started with Next-Generation Artifcial Intelligence through Reinforcement Learning, the MDP-driven Bellman equation relied on a reward matrix. In this chapter, we will get our hands dirty in a ...

Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.

Read now

Unlock full access

More than 5,000 organizations count on O’Reilly

O’Reilly covers everything we've got, with content to help us build a world-class technology community, upgrade the capabilities and competencies of our teams, and improve overall team performance as well as their engagement.

Julian F.

Head of Cybersecurity

I wanted to learn C and C++, but it didn't click for me until I picked up an O'Reilly book. When I went on the O’Reilly platform, I was astonished to find all the books there, plus live events and sandboxes so you could play around with the technology.

Addison B.

Field Engineer

I’ve been on the O’Reilly platform for more than eight years. I use a couple of learning platforms, but I'm on O'Reilly more than anybody else. When you're there, you start learning. I'm never disappointed.

Amir M.

Data Platform Tech Lead

I'm always learning. So when I got on to O'Reilly, I was like a kid in a candy store. There are playlists. There are answers. There's on-demand training. It's worth its weight in gold, in terms of what it allows me to do.

Mark W.

Embedded Software Engineer

Publisher Resources

ISBN: 9781839211539

Cloud Computing

Data Engineering

Data Science

AI & ML

Programming Languages

Software Architecture

IT/Ops

Security

Design

Business

Soft Skills

Artificial Intelligence By Example - Second Edition

by Denis Rothman

2

Building a Reward Matrix – Designing Your Datasets

Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.