51²è¹Ý

Representation Learning or so-called feature learning aims to train a machine to learn features and use them to perform specific tasks. Our research spans a range of ideas including building (1) generative models, (2) unsupervised / semi-supervised / supervised knowledge representation learning, (3) visualization and interpretation of learned representations, (4) integrated time series and structural learning, and (5) large-scale learning.

Social Aspects of Machine Learning Our faculty focuses on three aspects. The first is interactive ML which aims at keeping human in the loop, to leverage human interaction in a collaborative environment to facilitate machine learning applications. The second is security and privacy preserving ML, aiming to protect models and data through homomorphic encryption methods, data infrastructure and hybrid cloud. The third is artificial intelligence that makes use of gaming.

Machine Learning for Sciences Our faculty has expertise in applying machine/deep learning techniques to help solve real-world issues in the interdisciplinary fields including earth science, chemistry, medical and health sciences, and history. Some examples include behavioral analysis, pattern recognition, and anomaly detection over big data accumulated from mobile phones, remote sensors, facial recognition, and biomedical image/video inspection.

Service Oriented Software Engineering (SOSE) refers to the contemporary 3rd-generation software engineering, which focuses on leveraging reusable APIs to enable higher development production, faster to market, and less errors. Our faculty are specialized in leveraging machine/deep learning (ML/DL), data mining, natural language processing (NLP), and social networking analysis techniques to analyze past software behaviors and usage history in order to support: software search engine, software usage recommendation, semi-automatic big data analytics workflow generation, software prediction, requirements traceability, and cloud/edge hybrid resource management.

Software Testing evaluates and verifies that a software product or application does what it promises to do, and to prevent/reduce/tolerate bugs, improve performance, and reduce development costs. Our current areas of focus include reliability, usability, safety, and API testing.

Anomaly/defect/hazard/vulnerability detection/ resolution CS faculty focus on applying machine/deep learning, data mining, and natural language processing (NLP) techniques to conduct assessment through code/ behavioral/environmental analysis, in order to detect/ classify/resolve anomalies, defects, hazards, vulnerabilities, and malicious code early on.

Recommendation Our faculty are specialized in applying various machine/deep learning (ML/DL) techniques, such as recurrent neural networks, convolutional neural networks, and graph neural networks, to analyze software past behaviors and user interactions, in order to predict software usage trend and provide recommendations to software providers and cloud data center management.

Text Mining The expertise in CS is leveraged to analyze textual documents, by applying machine learning/deep learning (ML/DL), data mining, and natural language processing (NLP) techniques. Particularly, we train machine to read scientific papers, analyze company financial records, survey historical documents, examine government files, and inspect insurance contracts.

Education The research in CS aims to seamlessly integrate artificial intelligence, reinforcement learning, gaming, and distributed computing technologies to support and facilitate education.  Our faculty are currently working on creating massive labeled dataset to enable simulation and emulation of training environments, conducting active learning such as deep knowledge tracing, automating assessment by processing spoken language, increasing assessment efficiency, simulating museum activities, and creating an environment to enable interactions between students and video games.