Probability models uncertainty:
what you don’t know yet, but want to reason about.
Key takeaway:
Probability is a number between 0 and 1 that measures how likely an event is.
The whole topic is about defining events clearly and applying a few core rules consistently.
Probability quantifies uncertainty: a number between 0 and 1.
Probability often changes when we learn new information.
Conditional probability and Bayes’ theorem give a structured way to update beliefs using evidence.
Conditional probability updates probabilities after observing an event.
Bayes’ theorem lets you estimate a hidden cause from observed evidence.
Naïve Bayes turns Bayes’ theorem into a practical classifier by assuming conditional independence of features given the class.
flowchart TD
A[Conditional<br/>probability] -->|foundation| B[Bayes<br/>theorem]
D[Independent<br/>events] -->|implies| C[Independence]
C -->|simplifies| A
E[Prior] -->|with likelihood| B
F[Likelihood] -->|updates| H[Posterior]
G[Evidence] -->|normalises| B
B -->|yields| H
I[Naïve<br/>Bayes] -->|uses| B
J[Naïve<br/>assumption] -->|assumes| C
K[Features] -->|given class| J
L[Class] -->|conditions| J
I -->|predicts| M[Classification]
M -->|selects| L
style A fill:#90CAF9,stroke:#1E88E5,color:#000
style B fill:#90CAF9,stroke:#1E88E5,color:#000
style C fill:#90CAF9,stroke:#1E88E5,color:#000
style D fill:#CE93D8,stroke:#8E24AA,color:#000
style E fill:#CE93D8,stroke:#8E24AA,color:#000
style F fill:#CE93D8,stroke:#8E24AA,color:#000
style G fill:#CE93D8,stroke:#8E24AA,color:#000
style J fill:#CE93D8,stroke:#8E24AA,color:#000
style K fill:#CE93D8,stroke:#8E24AA,color:#000
style L fill:#CE93D8,stroke:#8E24AA,color:#000
style H fill:#C8E6C9,stroke:#2E7D32,color:#000
style I fill:#C8E6C9,stroke:#2E7D32,color:#000
style M fill:#C8E6C9,stroke:#2E7D32,color:#000
An intelligent agent connects perception with action. It observes an environment through sensors, uses the information it receives to decide what to do, and affects the environment through actuators.
This page develops the core vocabulary for reasoning about intelligent agents: percepts, actions, rationality, performance measures, PEAS and the different properties an environment can have.
Intelligent Agents: Notion of Agents and Environments, Rationality
flowchart TD
E["Environment"] -->|Percepts| S["Sensors"]
S --> A["Agent"]
A --> C["Actuators"]
C -->|Actions| E
style E fill:#E1F5FE
style S fill:#FFF9C4
style A fill:#C8E6C9
style C fill:#EDE7F6
Think of an agent as a continuous loop: observe → decide → act → observe again. An action may change the environment, so the next observation may be different.
This module applies parallel and distributed computing to ML algorithms. It examines how to divide computation, combine results, and balance communication against useful work during training.
Module coverage:
Parallel ML Algorithms: CNNs, gradient descent and SGD, SVM, k-means, k-nearest neighbours, decision trees, and random forests.
Distributed Training Strategies: data parallelism, model parallelism, pipeline parallelism, gradient checkpointing, and mixed-precision training.
Use M1: Foundations to interpret complexity and speedup. Continue to M3: Scale-Out Systems for the platforms and communication mechanisms that support larger training runs.
The Multi-Armed Bandit (MAB) problem is the simplest setting for studying decision-making under uncertainty.
An agent repeatedly chooses one of
\( k \)
actions. Each action produces a numerical reward drawn from an unknown distribution. The objective is to maximise the expected total reward over time.
The central challenge is deciding when to exploit current knowledge and when to explore uncertain alternatives.
Multi-armed Bandit Problem - Motivation and Problem Statement
Incremental solution to the stationary & non-stationary MAB problems
Exploration vs. Exploitation trade-off
Bandit Gradient Algorithm as Stochastic Gradient Ascent
Parallelisation divides computational work into parts that can execute concurrently. The purpose is to reduce completion time or increase throughput, but the gain depends on how much work is genuinely independent and how much overhead is introduced.
Course coverage:
Systems and performance: execution time, throughput, efficiency, and bottlenecks
Speedup—approaches and issues: Amdahl’s Law, serial work, overhead, and load imbalance
Data parallelism versus task parallelism versus request parallelism
Scale-out clusters: communication cost and its impact on speedup
The worked algorithms also show how divide-and-conquer and matrix operations expose parallel work.
