Introduction to Neural Network Modelling (PSYL10151)

Home / Psychology 3 / Introduction to Neural Network Modelling (PSYL10151)

Introduction to Neural Network Modelling (PSYL10151)

This course provides an introduction to neural networks and their use in understanding human and non-human animal cognition. In specific, students will be exposed to simple auto-associative, feed-forward, and recurrent network architectures, and Hebbian, back-propagation, and unsupervised training methods. Students will also be exposed to more recent developments in deep neural networks. The course emphasizes the use of neural networks as tools for understanding cognition and for instantiating cognitive theories.

Students will receive a crash course in the python programing language. They will use python to develop simple neural networks. Students will also be exposed to the keras library for building more complex neural networks. They will use keras to develop a larger network project.

Students will acquire skills in logical thinking, abstract thinking, mathematical and programmatic reasoning, and training in developing falsifiable psychological theories.

 

Lectures
Lectures 1-2: Python basics (reading: A byte of python, chaps. 7-14; suggest that you also read 15 and 16)
Lecture 3: Neural network basics and tools
Lecture 4: Hopfield nets (assignment 1 due)
Lecture 5: Perceptrons  
Lecture 6: Hebbian and error learning (assignment 2 due)
Lectures 7-8: Keras and lab
Lecture 9: Simple unsupervised learning (assignment 3 due)
Lectures 10: Building larger scale systems

 

Learning Outcomes: 

 

 Understanding of basic feed-forward, recurrent, and auto-associative neural networks and understanding of basic supervised and unsupervised methods for training neural networks.
 Appreciation of the scope of neural networks as tools for understanding cognition and solving problems.
 Understanding of the associations between the basic properties of neural networks, and our current understanding of real neural systems.
 Develop basic competence with the python programming language and the keras library.
 Students will acquire skills in logical thinking, abstract thinking, mathematical and programmatic reasoning, and training in developing falsifiable psychological theories.

 

Additional Information: 

 

Detailed Assessment Information - 

1) Students will complete three (3) assignments during the course for a total of 30% of the mark: 

Assignments 30%. Each assignment will include a programming component and a short write-up (500-1000 words) and each will contribute 10% towards the final mark. For all assignments, student code will be assessed on a pass/fail basis. For the first assignment, students will receive pass/fail credit with instructor feedback. For the second assignment, students will receive a mark for writing quality based on the feedback given for assignment 1. For assignment 3, students will provide peer feedback, and will be assessed on the level of feedback provided. 

2) Final assignment 70%. Students will complete a full network assignment including code and a complete write-up of the network description and the network results (2000 words). The final assignment will be marked according the University grading scheme.