Artificial Intelligence Cheat Sheet

This cheat sheet covers foundational Artificial Intelligence concepts, including search algorithms, knowledge representation, and machine learning techniques, crucial for understanding AI's goals and applications.

Core Principles

AI agents interact with environments using various architectures like simple reflex, model-based, goal-based, and learning agents.
Search algorithms like BFS and DFS are fundamental for problem-solving in AI, each with trade-offs in completeness, optimality, and time complexity.
Knowledge representation involves logic systems such as propositional logic, first-order logic, and probabilistic reasoning using Bayesian Networks.
Machine learning encompasses supervised, unsupervised, and reinforcement learning, with key concepts including overfitting, bias-variance tradeoff, and algorithms like k-NN, Decision Trees, Naive Bayes, and SVMs.

Action Steps

1. Define Artificial Intelligence: Explain its goals and major applications.
2. Differentiate rational agents from human thinking.
3. Explain agent-environment interaction with a diagram.
4. Categorize AI environments: fully/partially observable, deterministic/stochastic, episodic/sequential, static/dynamic, discrete/continuous.
5. Describe different agent architectures: simple reflex, model-based, goal-based, utility-based, and learning agents.
6. Formulate problems in AI and identify components of a well-defined search problem.
7. Compare Breadth-First Search (BFS) and Depth-First Search (DFS) on completeness, optimality, time complexity, and space complexity.
8. Compare Uniform Cost Search (UCS) and Iterative Deepening Search (IDS) on advantages and limitations.
9. Compare Greedy Best-First Search and A* Search on working and limitations.
10. Explain heuristic functions, admissibility, and consistency with examples.
11. Illustrate the Minimax algorithm for adversarial search with a game tree.
12. Explain Alpha-Beta pruning and its improvement over Minimax.
13. Describe knowledge-based agents, their architecture, and working.
14. Explain propositional logic and inference methods.
15. Differentiate propositional logic from First-Order Logic (FOL) with suitable examples.
16. Explain rule-based systems, their components, and forward/backward chaining.
17. Explain AI planning and the STRIPS representation.
18. Explain fuzzy logic, fuzzification, membership functions, and the difference from classical Boolean logic.
19. Explain probabilistic reasoning and Bayes' theorem with an example.
20. Define conditional independence and its importance in probabilistic reasoning and Bayesian Networks.
21. Explain the structure and working of a Bayesian Network with an example.
22. Define Machine Learning and explain supervised, unsupervised, and reinforcement learning with examples.
23. Differentiate between parametric and non-parametric learning models.
24. Explain the bias-variance tradeoff and its relation to overfitting and underfitting.
25. Explain techniques for regularization and prevention of overfitting.
26. Intuitively explain Gradient Descent, learning rate, cost function, and convergence.
27. Discuss k-Nearest Neighbors (k-NN): algorithm, choice of k, advantages, and limitations.
28. Describe Decision Trees, including splitting criteria like Entropy, Information Gain, and Gini Index.
29. Explain Naive Bayes classification and its assumption of conditional independence.
30. Explain Support Vector Machines (SVM), Artificial Neural Networks, Ensemble Methods/Boosting, EM/HMM, Q-Learning, and Responsible AI concepts.

Key Terms

Rational Agent: An agent that acts to achieve the best outcome or expected outcome, given its percept sequence and any built-in knowledge.
Breadth-First Search (BFS): A graph traversal algorithm that explores neighbor nodes first before moving to the next level, guaranteeing the shortest path in unweighted graphs.
Depth-First Search (DFS): A graph traversal algorithm that explores as far as possible along each branch before backtracking.
Heuristic Function: A function that estimates the cost from a given state to the goal state, used to guide search algorithms like Greedy Best-First Search and A*.
Minimax Algorithm: A decision-making algorithm used in AI for two-player, zero-sum games, aiming to minimize the possible loss for a worst-case (maximum) scenario.
Alpha-Beta Pruning: An optimization technique for the Minimax algorithm that reduces the number of nodes evaluated in the search tree.
Propositional Logic: A logic system dealing with propositions (statements that are either true or false) and logical connectives like AND, OR, NOT, IMPLIES.
First-Order Logic (FOL): A more expressive logic system than propositional logic, allowing the use of variables, quantifiers (like 'for all' and 'there exists'), and predicates.
Bayesian Network: A probabilistic graphical model that represents a set of random variables and their conditional dependencies using a directed acyclic graph (DAG).
Machine Learning: A field of AI that enables systems to learn from data and improve their performance on a task without being explicitly programmed.
Overfitting: A phenomenon in machine learning where a model learns the training data too well, including noise and outliers, leading to poor generalization on unseen data.
Bias-Variance Tradeoff: The relationship between two sources of error in machine learning: bias (simplifying assumptions) and variance (sensitivity to small fluctuations in the training set).
Gradient Descent: An iterative optimization algorithm used to find the minimum of a function by taking steps proportional to the negative of the gradient (or approximate gradient) of the function at a given point.
k-Nearest Neighbors (k-NN): A non-parametric, instance-based learning algorithm used for classification and regression, where a data point is classified based on the majority class of its 'k' nearest neighbors.
Decision Trees: A tree-like model of decisions and their possible consequences, used for classification and regression by recursively partitioning the data based on feature values.
Naive Bayes: A probabilistic classifier based on applying Bayes' theorem with a strong (naive) assumption of independence between features.
Support Vector Machine (SVM): A supervised learning model that finds an optimal hyperplane to separate data points into different classes, maximizing the margin between them.

Real World Examples

Game playing (e.g., Chess, Checkers): Demonstrates the application of search algorithms like Minimax with Alpha-Beta pruning to find optimal moves.
Spam filtering: Illustrates the use of Naive Bayes classification, assuming independence between words in an email.
Image recognition: Can be implemented using Support Vector Machines (SVMs) or Artificial Neural Networks to classify images based on learned features.
Medical diagnosis: Bayesian Networks can model dependencies between symptoms and diseases to aid in diagnosis.

Timeline

Module I: Foundations of AI, Search, Knowledge & Reasoning, AI & Intelligent Agents
Module II: Machine Learning & Responsible AI, Introduction to Machine Learning