The most exciting development in parallel computer architecture is the convergence of traditionally disparate approaches on a common machine structure. This book explains the forces behind this convergence of shared-memory, message-passing, data parallel, and data-driven computing architectures. It then examines the design issues that are critical to all parallel architecture across the full range of modern design, covering data access, communication performance, coordination of cooperative work, and correct implementation of useful semantics. It not only describes the hardware and software techniques for addressing each of these issues but also explores how these techniques interact in the same system. Examining architecture from an application-driven perspective, it provides comprehensive discussions of parallel programming for high performance and of workload-driven evaluation, based on understanding hardware-software interactions.
Nils J. Nilsson在其长期而卓越的研究生涯中为人工智能这门学科的发展作出了巨大的贡献。他的许多著作都堪称经典，如《机器学习》、《人工智能解题方法》、《人工智能的逻辑基础》、《人工智能原理》等。Nilsson博士是斯坦福大学终身教授，同时担任“人工智能和机器学习”的编委。历任斯坦福大学计算机系主任，SRI人工智能中心主任，美国人工智能学会主席等。
1 Introduction
1.1 What is AI
1.2 Approaches to Artificial Intelligence
1.3 Brief History of AI
1.4 Plan of the Book
1.5 Additional Readings and Discussion
I Reactive Machines
2 Stimulus-Response Agents
2.1 Perception and Action
2.1.1 Perception
2.1.2 Action
2.1.3 Boolean Algebra
2.1.4 Classes and Forms of Boolean Functions
2.2 Representing and Implementing Action Functions
2.2.1 Production Systems
2.2.2 Networks
2.2.3 The Subsumption Architecture
2.3 Additional Readings and Discussion
3 Neural Networks
3.1 Introduction
3.2 Training Single TLUs
3.2.1 TLU Geometry
3.2.2 Augmented Vectors
3.2.3 Gradient Descent Methods
3.2.4 The Widrow-Hoff Procedure
3.2.5 The Generalized Delta Procedure
3.2.6 The Error-Correction Procedure
3.3 Neural Networks
3.3.1 Motivation
3.3.2 Notation
3.3.3 The Backpropagation Method
3.3.4 Computing Weight Changes in the Final Layer
3.3.5 Computing Changes to the Weights in Intermediate Layers
3.4 Generalization, Accuracy, and Overfitting
3.5 Additional Readings and Discussion
4 Machine Evolution
4.1 Evolutionary Computation
4.2 Genetic Programming
4.2.1 Program Representation in GP
4.2.2 The GP Process
4.2.3 Evolving a Wall-Following Robot
4.3 Additional Readings and Discussion
5 State Machines
5.1 Representing the Environment by Feature Vectors
5.2 Elman Networks
5.3 Iconic Representations
5.4 Blackboard Systems
5.5 Additional Readings and Discussion
6 Robot Vision
6.1 Introduction
6.2 Steering a Van
6.3 Two Stages of Robot Vision
6.4 Image Processing
6.4.1 Averaging
6.4.2 Edge Enhancement
6.4.3 Combining Edge-Enhancement with Averaging
6.4.4 Region Finding
6.4.5 Using Image Attributes other than Intensity
6.5 Scene Analysis
6.5.1 Interpreting Lines and Curves in the Image
6.5.2 Model-Based Vision
6.6 Stereo Vision
6.7 Additional Readings and Discussion
II Search in State Spaces
7 Agents that Plan
7.1 Memory Versus Computation
7.2 State-Space Graphs
7.3 Searching Explicit State Spaces
7.4 Feature-Based State Spaces
7.5 Graph Notation 7.6 Additional Readings and Discussion
8 Uninformed Search
8.1 Formulating the State Space
8.2 Components of Implicit State-Space Graphs
8.3 Breadth-First Search
8.4 Depth-First or Bracktracking Search
8.5 Iterative Deepening
8.6 Additional Readings and Discussion
9 Heuristic Search
9.1 Using Evaluation Functions
9.2 A General Graph-Searching Algorithm
9.2.1 Algorithm A
9.2.2 Admissibility of A
9.2.3 The Consistency (or Monotone) Condition
9.2.4 Iterative-Deepening A
9.2.5 Recursive Best-First Search
9.3 Heuristic Functions and Search Efficiency
9.4 Additional Readings and Discussion
10 Planning, Acting, and Learning
10.1 The Sense/Plan/Act Cycle
10.2 Approximate Search
10.2.1 Island-Driven Search
10.2.2 Hierarchical Search
10.2.3 Limited-Horizon Search
10.2.4 Cycles
10.2.5 Building Reactive Procedures
10.3 Learning Heuristic Functions
10.3.1 Explicit Graphs
10.3.2 Implicit Graphs
10.4 Rewards Instead of Goals
10.5 Additional Readings and Discussion
11 Alternative Search Formulations and Applications
11.1 Assignment Problems
11.2 Constructive Methods
11.3 Heuristic Repair
11.4 Function Optimization
12 Adversarial Search
12.1 Two-Agent Games
12.2 The Minimax Procedure
12.3 The Alpha-Beta Procedure
12.4 The Search Efficiency of the Alpha-Beta Procedure
12.5 Other Important Matters
12.6 Games of Chance
12.7 Learning Evaluation Functions
12.8 Additional Readings and Discussion
III Knowledge Representation and Reasoning
13 The Propositional Calculus
13.1 Using Constraints on Feature Values
13.2 The Language
13.3 Rules of Inference
13.4 Definition of Proof
13.5 Semantics
13.5.1 Interpretations
13.5.2 The Propositional Truth Table
13.5.3 Satisfiability and Models
13.5.4 Validity
13.5.5 Equivalence
13.5.6 Entailment
13.6 Soundness and Completeness
13.7 The PSAT Problem
13.8 Other Important Topics
13.8.1 Language Distinctions
13.8.2 Metatheorems
13.8.3 Associative Laws
13.8.4 Distributive Laws
14 Resolution in The Propositional Calculus
14.1 A New Rule of Inference: Resolution
14.1.1 Clauses as wwf
14.1.2 Resolution on Clauses
14.1.3 Soundness of Resolution
14.2 Converting Arbitrary wffs to Conjunctions of Clauses
14.3 Resolution Refutations
14.4 Resolution Refutation Search Strategies
14.4.1 Ordering Strategies
14.4.2 Refinement Strategies
14.5 Horn Clauses
15 The Predicate Calculus
15.1 Motivation
15.2 The Language and its Syntax
15.3 Semantics
15.3.1 Worlds
15.3.2 Interpretations
15.3.3 Models and Related Notions
15.3.4 Knowledge
15.4 Quantification
15.5 Semantics of Quantifiers
15.5.1 Universal Quantifiers
15.5.2 Existential Quantifiers
15.5.3 Useful Equivalences
15.5.4 Rules of Inference
15.6 Predicate Calculus as a Language for Representing Knowledge
15.6.1 Conceptualizations
15.6.2 Examples
15.7 Additional Readings and Discussion
16 Resolution in the Predicate Calculus
16.1 Unification
16.2 Predicate-Calculus Resolution
16.3 Completeness and Soundness
16.4 Converting Arbitrary wffs to Clause Form
16.5 Using Resolution to Prove Theorems
16.6 Answer Extraction
16.7 The Equality Predicate
16.8 Additional Readings and Discussion
17 Knowledge-Based Systems
17.1 Confronting the Real World
17.2 Reasoning Using Horn Clauses
17.3 Maintenance in Dynamic Knowledge Bases
17.4 Rule-Based Expert Systems
17.5 Rule Learning
17.5.1 Learning Propositional Calculus Rules
17.5.2 Learning First-Order Logic Rules
17.5.3 Explanation-Based Generalization
17.6 Additional Readings and Discussion
18 Representing Commonsense Knowledge
18.1 The Commonsense World
18.1.1 What is Commonsense Knowledge
18.1.2 Difficulties in Representing Commonsense Knowledge
18.1.3 The Importance of Commonsense Knowledge
18.1.4 Research Areas
18.2 Time
18.3 Knowledge Representation by Networks
18.3.1 Taxonomic Knowledge
18.3.2 Semantic Networks
18.3.3 Nonmonotonic Reasoning in Semantic Networks
18.3.4 Frames
18.4 Additional Readings and Discussion
19 Reasoning with Uncertain Information
19.1 Review of Probability Theory
19.1.1 Fundamental Ideas
19.1.2 Conditional Probabilities
19.2 Probabilistic Inference
19.2.1 A General Method
19.2.2 Conditional Independence
19.3 Bayes Networks
19.4 Patterns of Inference in Bayes Networks
19.5 Uncertain Evidence
19.6 D-Seperation
19.7 Probabilistic Inference in Polytrees
19.7.1 Evidence Above
19.7.2 Evidence Below
19.7.3 Evidence Above and Below
19.7.4 A Numerical Example
19.8 Additional Readings and Discussion
20 Learning and Acting with Bayes Nets
20.1 Learning Bayes Nets
20.1.1 Known Network Structure
20.1.2 Learning Network Structure
20.2 Probabilistic Inference and Action
20.2.1 The General Setting
20.2.2 An Extended Example
20.2.3 Generalizing the Example
20.3 Additional Readings and Discussion
IV Planning Method Based on Logic
21 The Situation Calculus
21.1 Reasoning about States and Actions
21.2 Some Difficulties
21.2.1 Frame Axioms
21.2.2 Qualifications
21.2.3 Ramifications
21.3 Generating Plans
21.4 Additional Reading and Discussion
22 Planning
22.1 STRIPS Planning Systems
22.1.1 Describing States and Goals
22.1.2 Forward Search Methods
22.1.3 Recursive STRIPS
22.1.4 Plans with Runtime Conditionals
22.1.5 The Sussman Anomaly
22.1.6 Backward Search Methods
22.2 Plan Spaces and Partial-Order Planning
22.3 Hierarchical Planning
22.3.1 ABSTRIPS
22.3.2 Combining Hierarchical and Partial-Order Planning
22.4 Learning Plans'
22.5 Additional Readings and Discussion
V Communication and Integration
23 Multiple Agents
23.1 Interacting Agents
23.2 Models of Other Agents
23.2.1 Varieties of Models
23.2.2 Simulation Strategies
23.2.3 Simulated Databases
23.2.4 The Intentional Stance
23.3 A Modal Logic of Knowledge
23.3.1 Modal Operators
23.3.2 Knowledge Axioms
23.3.3 Reasoning about Other Agents' Knowledge
23.3.4 Predicting Actions of Other Agents
23.4 Additional Readings and Discussion
24 Communication Among Agents
24.1 Speech Acts
24.1.1 Planning Speech Acts
24.1.2 Implementing Speech Acts
24.2 Understanding Language Strings
24.2.1 Phrase-Structure Grammars
24.2.2 Semantic Analysis
24.2.3 Expanding the Grammar
24.3 Efficient Communication
24.3.1 Use of Context
24.3.2 Use of Knowledge to Resolve Ambiguities
24.4 Natural Language Processing
24.5 Additional Readings and Discussion
25 Agent Architectures
25.1 Three-Level Architectures
25.2 Goal Arbitration
25.3 The Triple-Tower Architecture
25.4 Bootstrapping
25.5 Additional Readings and Discussion
