Untitled Deck Flashcards
What are the four approaches to defining Artificial Intelligence (AI)?
Acting humanly, Thinking humanly, Thinking rationally, Acting rationally.
What is the Turing Test?
A test to determine if a machine can exhibit intelligent behavior indistinguishable from that of a human.
What is the difference between thinking humanly and thinking rationally in AI?
Thinking humanly focuses on modeling the human brain and behavior (cognitive science), while thinking rationally focuses on logical reasoning (e.g., Aristotle’s syllogisms).
Define an intelligent agent.
An entity that perceives its environment through sensors and acts upon it using actuators to achieve its goals.
List the properties of task environments for agents.
Observable (fully/partially), Single/multi-agent, Deterministic/stochastic, Episodic/sequential, Static/dynamic, Discrete/continuous, Known/unknown.
What was the significance of the Dartmouth Workshop (1956)?
It marked the birth of AI as a field of study.
What is an ‘AI Winter’?
A period of reduced funding and interest in AI due to overhyped promises and underwhelming results.
What technological advancement reignited interest in AI in the 2000s?
The emergence of Big Data and Deep Learning powered by GPUs.
What are the four types of intelligent agents?
Rule-based (reactive), Model-based, Goal-based, Utility-based.
What is the role of a ‘critic’ in learning agents?
To provide feedback to the agent to improve its performance.
What does ‘autonomous learning’ mean for intelligent agents?
The agent’s behavior is determined by its own experience, not preprogrammed instructions.
What is supervised learning?
Learning a function from labeled training data to predict outputs for unseen inputs.
What is Occam’s Razor in machine learning?
The principle that simpler models are preferred as they generalize better.
Define a probabilistic classifier.
A classifier that outputs the probabilities of a sample belonging to each class.
What is binary classification?
A classification task where there are only two possible class labels.
What is a confusion matrix?
A matrix that summarizes the performance of a classification model by showing the counts of true positives, true negatives, false positives, and false negatives.
How is accuracy calculated?
Accuracy = (TP + TN) / (TP + TN + FP + FN).
What does an ROC curve represent?
The trade-off between true positive rate and false positive rate for different classification thresholds.
What is cross-validation?
A technique for evaluating a model’s performance by training and testing it on different subsets of the data.
What is the main assumption of Naïve Bayes classifiers?
All input features are conditionally independent given the class label.
How is the posterior probability estimated in Naïve Bayes?
Using the m-estimate: P(x|y) ≈ (n’ + mp) / (n + m).
What is an example of a use case for Naïve Bayes classifiers?
Spam email detection.
What is a decision tree?
A tree structure where internal nodes represent tests on attributes, branches represent outcomes, and leaves represent class labels or decisions.
What type of learning task does a decision tree address?
Supervised learning for classification and regression.
What is the purpose of propositional logic in decision trees?
Each path from root to leaf represents a rule in propositional logic.
What is entropy in the context of decision trees?
A measure of uncertainty or impurity in the dataset.
What metric does the ID3 algorithm use to select the best attribute?
Information Gain (IG).
How is entropy calculated in ID3?
H(Examples) = ∑ −P(v) ⋅ log2(P(v)), where P(v) is the probability of class v.
What are common drawbacks of the ID3 algorithm?
Overfitting, local optima, and bias towards shorter trees.
What is information gain?
The reduction in entropy achieved by partitioning the examples based on an attribute.
What are two methods to reduce overfitting in decision trees?
Early stopping and pruning.
What is the role of validation sets in pruning?
To identify and remove irrelevant parts of the tree post-training.
What is a ‘hill-climbing’ search in ID3?
A search strategy that selects the best immediate option without considering global optimization.
