James Mountstephens

Game AI generation using evolutionary multi-objective optimization

This paper presents the design and evaluation of a full AI controller for Real-Time Strategy (RTS) games using techniques from Evolutionary Computing (EC). The design is novel in its use of a modified Pareto Differential Evolution (PDE) algorithm for bi-objective optimization of the weights of an Artificial Neural Network (ANN) controller when only single-objective optimization has so far been studied. The two main aims of this research are to: (1) develop controllers capable of defeating opponents of varying difficulty levels, which may assist in commercial RTS AI development, and (2) minimize the number of neurons used in the ANN architecture, an issue primarily of efficiency. Experimental results using the popular Warcraft III platform demonstrate success with both aims: the optimized controller was able to win any battle using only a minimal number of hidden neurons, but sub-optimal controllers were able to provide opponents of any intermediate difficulty.

Deep learning for EEG-Based preference classification

Electroencephalogram (EEG)-based emotion classification is rapidly becoming one of the most intensely studied areas of brain-computer interfacing (BCI). The ability to passively identify yet accurately correlate brainwaves with our immediate emotions opens up truly meaningful and previously unattainable human-computer interactions such as in forensic neuroscience, rehabilitative medicine, affective entertainment and neuro-marketing. One particularly useful yet rarely explored areas of EEG-based emotion classification is preference recognition [1], which is simply the detection of like versus dislike. Within the limited investigations into preference classification, all reported studies were based on musically-induced stimuli except for a single study which used 2D images. The main objective of this study is to apply deep learning, which has been shown to produce state-of-the-art results in diverse hard problems such as in computer vision, natural language processing and audio recognition, to 3D object pre...

Using Biologically-Inspired Visual Features To Model The Restorative Potential Of Scenes

This paper describes novel, interdisciplinary work towards learning the properties of visual scenes that restore our directed attention from fatigue. A groundtruth dataset of images rated for restorative potential was constructed and validated using human subjects, and biologically-inspired image features were used to train a number of regression models for this rating. The trained models were used to predict the restorative potential of unseen images and the predictions were tested using human subjects, with promising results.