EXPLORATION INTO AI

2021 SEMESTER 1 PROJECT  [Compsys726]

Completed.

COMPSYS 726 - Robotics and Intelligent Systems aims to teach students about robot manipulators, mobile robots, navigation techniques, planning and programming of robot actions, sensors and actuators, kinematic analysis and includes topics in artificial intelligence, artificial neural nets, fuzzy systems, genetic algorithms. This project specifically focused on understanding the key elements which make up an artificial neural network. By generating the various activation functions from scratch and quantifying their performances, a better understanding of ANNs can be gained.

FunctionTiming1.png

LAB 1 - Forward and derivative mappings of various activation functions

Photo: Activation Function Timings.

[REPORT]

Lab 2 - Writing a two-layer inference engine for an ANN

Photo: Confusion Matrix plot.

[REPORT]

0.9.png

lab 3 - Implementation of two gradient descent algorithms

Photo: Gradient Descent Visualized.

[REPORT]

Sigmoid_iris_mseLRplot.png

Technical review - Is a specific morphology of an Activation Function important?

Photo: Effect of a varying learning rate on Mean Squared Error.

[REPORT]

Previous
Previous

TRACK

Next
Next

BONE SCREWS