Danny Silver's Thesis and Projects page:
for details, email danny.silver@acadiau.ca
My primary research interest is Machine Learning, a sub-area of Artificial Intelligence (AI).
Machine Learning has become one of the most respected areas of AI being used commercially in
Data Analytics, Data Mining, Intelligent Agents and Adaptive Systems.
Specifically, I am interested in Lifelong Machine Learning (LML) systems and most recently
such systems that can Learn to Reason.
I also have interest in the following research areas: user modeling and adaptive user interfaces,
handheld and wireless technology, robotics, knowledge management and E-Commerce.
I invite you to become a research associate with the
Machine Learning Research Laboratory, located in CAR 112.
Suggested projects and thesis topics are listed below:
Machine Learning Theory
-
As part of my research into LML here are various projects related to the sequentially learning of
tasks over a software agent’s or robot’s lifetime. Such systems must be capable of learning and
retaining task knowledge for use when later learning another related task.
-
We are using unsupervised and supervised deep learning methods to create LML and multimodal
learning systems – for an example see https://ml3cpu.acadiau.ca/
-
We work with deep learning systems such as TensorFlow and Keras – see
https://www.tensorflow.org/
and
https://mxnet.incubator.apache.org/
and
https://keras.io/.
-
We are interested in Learning to Reason and have one the first web repositories dedicated to
Lifelong Machine Learning and Reasoning at http://lmlr.acadiau.ca/.
The goal over the next five years is to develop a system that can reason (at least in some simple way)
from what it has learned from examples. This combines expertise in machine learning with knowledge
representation and reasoning. We are interested in developing a system that can tell us what
function it is currently executing (the reasoning).
Machine Learning Applications project ideas
-
Develop a system that can identify one or more objects in an image.
-
Follow-up prior work on a neural network that can morphe an image based on the sounds that it hears.
-
Design and develop an interesting interactive game based on the use of machine learning technology;
eg. agents that learn to compete against each other.
-
Design and develop a real or virtual robotic system that learns to maneuver through a maze using
reinforcement learning.
-
Develop a system that can morph images (such as human faces) based on machine learning technology
(from sad to happy).
-
Develop an associative memory system that can recall a complete image from a partial or noisy image.
Data Analytics and Data Mining
-
Develop and test a focused, application specific, data mining system using a machine learning algorithm;
examples are:
-
Create a system for use by environmental studies or by the biology department.
-
Create a data mining tool for a problem in kinesiology or recreation.
-
Apply a data mining / machine learning system to a complex phenomenon in science, engineering,
health care, or business; examples are:
-
Development of a predictive model for rainfall, plant growth, population changes
-
Collection of experimental data and modeling of a physical or chemical response to varying parameters
-
Conduct a data mining study using health care data from a collection of patient information
-
Develop explanatory or predictive models using GIS and business data.