book

Grokking Artificial Intelligence Algorithms

by Rishal Hurbans

August 2020

Intermediate to advanced

392 pages

9h 54m

English

Manning Publications

Read now

Unlock full access

Our obsession with technology and automationEthics, legal matters, and our responsibilityIntention and impact: Understanding your vision and goalsUnintended use: Protecting against malicious useUnintended bias: Building solutions for everyoneThe law, privacy, and consent: Knowing the importance of core valuesSingularity: Exploring the unknownacknowledgmentsabout this bookWho should read this bookHow this book is organized: A roadmapAbout the CodeliveBook discussion forumOther online resourcesabout the author
What is artificial intelligence?Defining AIUnderstanding that data is core to AI algorithmsViewing algorithms as instructions in recipesA brief history of artificial intelligenceProblem types and problem-solving paradigmsSearch problems: Find a path to a solutionOptimization problems: Find a good solutionPrediction and classification problems: Learn from patterns in dataClustering problems: Identify patterns in dataDeterministic models: Same result each time it’s calculatedStochastic/probabilistic models: Potentially different result each time it’s calculatedIntuition of artificial intelligence conceptsNarrow intelligence: Specific-purpose solutionsGeneral intelligence: Humanlike solutionsSuper intelligence: The great unknownOld AI and new AISearch algorithmsBiology-inspired algorithmsMachine learning algorithmsDeep learning algorithmsUses for artificial intelligence algorithmsAgriculture: Optimal plant growthBanking: Fraud detectionCybersecurity: Attack detection and handlingHealth care: Diagnosis of patientsLogistics: Routing and optimizationTelecoms: Optimizing networksGames: Creating AI agentsArt: Creating masterpiecesSummary of Intuition of artificial intelligence
What are planning and searching?Cost of computation: The reason for smart algorithmsProblems applicable to searching algorithmsRepresenting state: Creating a framework to represent problem spaces and solutionsGraphs: Representing search problems and solutionsRepresenting a graph as a concrete data structureExercise: Represent a Graph as a MatrixSolution: Represent a Graph as a MatrixTrees: The concrete structures used to represent search solutionsUninformed search: Looking blindly for solutionsBreadth-first search: Looking wide before looking deepExercise: Determine the path to the solutionSolution: Determine the path to the solutionDepth-first search: Looking deep before looking wideExercise: Determine the path to the solutionSolution: Determine the path to the solutionUse cases for uninformed search algorithmsOptional: More about graph categoriesOptional: More ways to represent graphsIncidence matrixAdjacency listSummary of search fundamentals
Defining heuristics: Designing educated guessesThought Experiment: Given the following scenario, what heuristic can you imagine?Thought Experiment: Possible solutionInformed search: Looking for solutions with guidanceA* searchUse cases for informed search algorithmsAdversarial search: Looking for solutions in a changing environmentA simple adversarial problemMin-max search: Simulate actions and choose the best futureHeuristicsExercise: What values would propagate in the following Min-max tree?Solution: What values would propagate in the following Min-max tree?Alpha-beta pruning: Optimize by exploring the sensible paths onlyUse cases for adversarial search algorithmsSummary of Intelligent search
What is evolution?Problems applicable to evolutionary algorithmsGenetic algorithm: Life cycleEncoding the solution spacesBinary encoding: Representing possible solutions with zeros and onesExercise: What is a possible encoding for the following problem?Solution: What is a possible encoding for the following problem?Creating a population of solutionsMeasuring fitness of individuals in a populationSelecting parents based on their fitnessSteady state: Replacing a portion of the population each generationGenerational: Replacing the entire population each generationRoulette wheel: Selecting parents and surviving individualsReproducing individuals from parentsSingle-point crossover: Inheriting one part from each parentTwo-point crossover: Inheriting more parts from each parentUniform crossover: Inheriting many parts from each parentExercise: What outcome would uniform crossover generate for these chromosomes?Solution: What outcome would uniform crossover generate for these chromosomes?Bit-string mutation for binary encodingFlip-bit mutation for binary encodingPopulating the next generationExploration vs. exploitationStopping conditionsConfiguring the parameters of a genetic algorithmUse cases for evolutionary algorithmsSummary of evolutionary algorithms
Evolutionary algorithm life cycleAlternative selection strategiesRank selection: Even the playing fieldTournament selection: Let them fightElitism selection: Choose only the bestReal-value encoding: Working with real numbersReal-value encoding at its coreArithmetic crossover: Reproduce with mathBoundary mutationArithmetic mutationOrder encoding: Working with sequencesImportance of the fitness functionOrder encoding at its coreOrder mutation: Order/permutation encodingTree encoding: Working with hierarchiesTree encoding at its coreTree crossover: Inheriting portions of a treeChange node mutation: Changing the value of a nodeCommon types of evolutionary algorithmsGenetic programmingEvolutionary programmingGlossary of evolutionary algorithm termsMore use cases for evolutionary algorithmsSummary of advanced evolutionary approaches
What is swarm intelligence?Problems applicable to ant colony optimizationExercise: Find the shortest path in this carnival configuration by handSolution: Find the shortest path in this carnival configuration by handRepresenting state: What do paths and ants look like?The ant colony optimization algorithm life cycleInitialize the pheromone trailsSet up the population of antsChoose the next visit for each antThe stochastic nature of antsSelecting destination based on a heuristicExercise: Determine the probabilities of visiting the attractions with the following informationSolution: Determine the probabilities of visiting the attractions with the following informationUpdate the pheromone trailsUpdating pheromones due to evaporationUpdating pheromones based on ant toursExercise: Calculate the pheromone update given the following scenarioSolution: Calculate the pheromone update given the following scenarioUpdate the best solutionDetermine the stopping criteriaUse cases for ant colony optimization algorithmsSummary of ant colony optimization

What is particle swarm optimization?Optimization problems: A slightly more technical perspectiveExercise: How many dimensions will the search space for the following scenario be?Solution: How many dimensions will the search space for the following scenario be?Problems applicable to particle swarm optimizationRepresenting state: What do particles look like?Particle swarm optimization life cycleInitialize the population of particlesCalculate the fitness of each particleExercise: What would the fitness be for the following inputs given the drone fitness function?Solution: What would the fitness be for the following inputs given the drone fitness function?Update the position of each particleThe components of updating velocityUpdating velocityPosition updateExercise: Calculate the new velocity and position for particle 1 given the following information about the particlesSolution: Calculate the new velocity and position for particle 1 given the following information about the particlesDetermine the stopping criteriaUse cases for particle swarm optimization algorithmsSummary of particle swarm optimization
What is machine learning?Problems applicable to machine learningSupervised learningUnsupervised learningReinforcement learningA machine learning workflowCollecting and understanding data: Know your contextPreparing data: Clean and wrangleMissing dataAmbiguous valuesEncoding categorical dataExercise: Identify and fix the problem data in this exampleSolution: Identify and fix the problem data in this exampleTesting and training dataTraining a model: Predict with linear regressionFitting a line to the dataFinding the mean of the featuresFinding regression lines with the least-squares methodExercise: Calculate a regression line using the least-squares methodSolution: Calculate a regression line using the least-squares methodTesting the model: Determine the accuracy of the modelSeparating training and testing dataMeasuring the performance of the lineImproving accuracyClassification with decision treesClassification problems: Either this or thatExercise: Regression vs. ClassificationSolution: Regression vs. ClassificationThe basics of decision treesTraining decision treesData structures for decision treesDecision-tree learning life cycleExercise: Calculating uncertainty and information gain for a questionSolution: Calculating uncertainty and information gain for a questionClassifying examples with decision treesOther popular machine learning algorithmsUse cases for machine learning algorithmsSummary of machine learning
What are artificial neural networks?The Perceptron: A representation of a neuronExercise: Calculate the output of the following input for the PerceptronSolution: Calculate the output of the following input for the PerceptronDefining artificial neural networksForward propagation: Using a trained ANNExercise: Calculate the prediction for the example BY using forward propagation with the following ANNSolution: Calculate the prediction for the example BY using forward propagation with the following ANNBackpropagation: Training an ANNPhase A: SetupPhase B: Forward propagationPhase C: TrainingExercise: Calculate the new weights for the highlighted weightsSolution: Calculate the new weights for the highlighted weightsOptions for activation functionsDesigning artificial neural networksInputs and outputsHidden layers and nodesWeightsBiasActivation functionsCost function and learning rateArtificial neural network types and use casesConvolutional neural networkRecurrent neural networkGenerative adversarial networkSummary of artificial neural networks
What is reinforcement learning?The inspiration for reinforcement learningProblems applicable to reinforcement learningThe life cycle of reinforcement learningSimulation and data: Make the environment come aliveTraining with the simulation using Q-learningExercise: Calculate the change in values for the Q-tableSolution: Calculate the change in values for the Q-tableTesting with the simulation and Q-tableMeasuring the performance of trainingModel-free and model-based learningDeep learning approaches to reinforcement learningUse cases for reinforcement learningRoboticsRecommendation enginesFinancial tradingGame playingSummary of reinforcement learning

Content preview from Grokking Artificial Intelligence Algorithms

2 Search fundamentals

This chapter covers

The intuition of planning and searching
Identifying problems suited to be solved using search algorithms
Representing problem spaces in a way suitable to be processed by search algorithms
Understanding and designing fundamental search algorithms to solve problems

What are planning and searching?

When we think about what makes us intelligent, the ability to plan before carrying out actions is a prominent attribute. Before embarking on a trip to a different country, before starting a new project, before writing functions in code, planning happens. Planning happens at different levels of detail in different contexts to strive for the best possible outcome when carrying out the tasks involved ...