Artificial Intelligence and Learning Machines

Learning Outcomes

Subject Specific Intellectual and Research Skills

Having successfully completed this module you will be able to:

• Choose and justify approaches and representations for solving particular tasks
• Derive the action or prediction common AI algorithms will make given particular knowledge, problems or states

Knowledge and Understanding

Having successfully completed this module, you will be able to demonstrate knowledge and understanding of:

• Demonstrate knowledge of the broad ranges of approaches to AI their their classification
• Understand the achievements and limitations of different approaches to AI
• Demonstrate awareness of the philosophical bases of artificial intelligence
• Discuss the responsibilities of AI engineers with respect to ethical aspects of learning technologies

Subject Specific Practical Skills

Having successfully completed this module you will be able to:

• Gain facility in performing first-order reasoning over knowledge bases using logic/relation programming languages (such as Python Kanren)
• Construct and make predictions with simple classification or regression trees (e.g. using CART) and mine some simple association rules
• Gain facility in creating and comparing simple game playing AIs (e.g. tic-tac-toe with game-trees and RL)

What is intelligence and what is artificial intelligence?

• Tests of AI: The Turing test. Searle's Chinese room argument.
• Classifications of AI: strong vs weak, symbolic vs sub-symbolic, etc.
• Fundamental problems in AI - e.g. grounding

Representation of data

• Symbolic representations
• Sub-symbolic representations

* Distributed representations

• Abstraction of representations

* Representational hierarchies, taxonomies

Representation of knowledge

• Parameters in functional expressions, algorithms and programs
• Decision trees
• Formal grammars
• Production rules
• Formal logic (predicate logic; propositional logic; first order reasoning)
• Frames and schemas
• Taxonomies
• Graphs, networks & ontologies (examples e.g. Wordnet and Cyc)

AI as learning

• Learning from examples

* KNN-classification and regression

* Learning rules from examples; inductive learning; simple decision tree learning; association rule mining

• Learning by correcting mistakes

* Early statistical learning machines - e.g. The MK1 Perceptron

* Linear classifiers

* Neural networks

* Differentiable programming

• Playing games and Learning by playing games

* Early game AIs (e.g. Strachey's "M. U. C. Draughts", Arther Samuel's Draughts, etc)

* Game trees, Tree search (Alpha–beta pruning, monte-carlo tree search)

* Learning state=>action pairs using reinforcement learning

Learning by being told & reasoning (GOFAI)

• Solving problems by searching over representations

* Finding satisfactory paths (e.g. depth-first and breadth-first, iterative deepening, local search and heuristic search).

* Finding optimal paths (e.g. branch and bound, dynamic programming, A*).

Inherent problems in AI

• Biases in data and knowledge representation
• Fragility of symbolic approaches
• Unexplainability of sub-symbolic approaches

The module consists of:

• Lectures
• Combined tutorials and computing laboratory sessions

Summative

This is how we’ll formally assess what you have learned in this module.

Referral

This is how we’ll assess you if you don’t meet the criteria to pass this module.

Repeat

An internal repeat is where you take all of your modules again, including any you passed. An external repeat is where you only re-take the modules you failed.