RESEARCH
Featured Current Research Projects
CMTE student works in collaboration on Machine Learning-based multivariate time series anomaly detection
Hierarchical Reinforcement Learning
Our work on Hierarchical Reinforcement Learning has been accepted to NeurIPS 2020 as part of the Deep Reinforcement Learning Workshop. We proposed two novel agents which improve the performance of existing methods in the case of deep and multi-agent reinforcement learning. Our first algorithm, the Evolution-based Soft Actor Critic (ESAC) makes use of evolutionary computing and automatic tuning methods to improve scalable control. The method is suitable for high-dimensional tasks such as robotic control and high-frequency trading. Our second algorithm, the Energy-based MIXer (EMIX) proposes a novel surprise minimization scheme which makes multiple agents robust to sudden changes in the environment. The method is found to be robust in large number of agents and market fluctuations. If you would like to know more about our work then check out our papers-
Francis investigated the anomaly detection problem on multivariate time series data. In particular, he studied two different directions for solving this problem: the point-based approach and the range-based approach. For the point-based approach, one novel Transformer-based model: Transformer Conditional Variational Autoencoder (T-CVAE) has been designed and compared with state-of-the-art multivariate time series anomaly detection baselines. Through exhaustive experiments on three open-source datasets, our model T-CVAE has outperformed all baseline models in terms of the F1 score. On the other hand, for the range-based approach, we have developed and implemented two time series encoding techniques: Outer product Matrix (OM) and Gramian Angular Field Matrix (GAFM).We have compared the two methods with the existing Gram Matrix approach in the literature. Based on empirical experiment results, we have found that GAFM time series encoding performs best among the three in terms of the F1 measurement.
Ongoing Current Research Projects
01
Utilizing Reinforcement Learning for Enhanced Trading Strategies
02
Utilizing Graph-Based Models for Index Tracking
03
Biometric Recognition Using Response to Acoustic Stimulation
04
Dynamic Malicious Code Detection Using Machine Learning
05
Development of a Recommendation Engine for Online B2B Customers
06
Using NLP to Detect Malware
07
Using Machine Learning for Workforce Optimization
08
