DALT7011 Introduction to Machine.

DALT7011 Introduction to Machine - Oxford Brookes University

Image Classification

Learning Outcome 1. Evaluate and articulate the issues and challenges in machine learning, including model selection, complexity and feature selection

Learning Outcome 2. Demonstrate a working knowledge of the variety of mathematical techniques normally adopted for machine learning problems, and of their application to creating effective solutions

Learning Outcome 3. Critically evaluate the performance and drawbacks of a proposed solution to a machine learning problem

Learning Outcome 4. Create solutions to machine learning problems using appropriate software

Report structure and assessment

1) Introduction

1) Explain the general notion of a classification problem in machine learning, and how CIFAR-10 formally fits into that framework.

2) Explain how the generalization error of classification algorithms can be robustly measured on datasets, and justify how you are going to use the methods in the remainder of the coursework.

2) Realize and describe an experiment in R that applies clustering to the dataset. Investigate how well the clusters separate the ten different classes. Assume the clusters are used for supervised classification, and a new data point is given a label based on the majority label of training in that cluster. How many clusters are needed to reach 25%, 50%, 75% classification accuracy?

3) Realize and describe an experiment in R that evaluates the classification error for k-nearest-neighbors (kNN) with PCA pre-processing on the CIFAR-10 dataset. Determine the most suitable value for the dimension and neighbourhood size experimentally with a suitable error measurement. Use appropriate illustrations and diagrams as well as statistics.