Explainable AI + VIS
Human-Centered AI with the Power of Visualization
About
With the recent advances in machine learning, especially deep learning, we have witnessed increasing applications of AI in various domains. The past few years have seen a growing interest towards more explainable machine learning systems.
At HKUST VisLab, we focus on human-centered approaches to make AI explainable, interactive, and trusworthy. In critical domains such as finance, security, and healthcare, explanability allows us to make more reliable decisions powered by the collaboration of machine- and human-intelligence. Interactivity makes the exploration, creation, and customization of AI at ease. With XAI, it is easier to build trustworthy AI solutions: we trust its performance, know when and where it is possible to fail, and can decide whether the AI solution meets our needs.
Publications
GNNLens: A Visual Analytics Approach for Prediction Error Diagnosis of Graph Neural Networks
AQX: Explaining Air Quality Forecast for Verifying Domain Knowledge using Feature Importance Visualization
VBridge: Connecting the Dots Between Features, Explanations, and Data for Healthcare Models
M2Lens: Visualizing and Explaining Multimodal Models for Sentiment Analysis
Visual Analysis of Discrimination in Machine Learning
HypoML: Visual Analysis for Hypothesis-based Evaluation of Machine Learning Models
DECE: Decision Explorer with Counterfactual Explanations for Machine Learning Models
DFSeer: A Visual Analytics Approach to Facilitate Model Selection for Demand Forecasting
ProtoSteer: Steering Deep Sequence Model with Prototypes
Visual Genealogy of Deep Neural Networks
Interpretable and Steerable Sequence Learning via Prototypes
ATMSeer: Increasing Transparency and Controllability in Automated Machine Learning
RuleMatrix: Visualizing and Understanding Classifiers with Rules
DeepTracker: Visualizing the Training Process of Convolutional Neural Networks
iForest: Interpreting Random Forests via Visual Analytics
EmbeddingVis: A Visual Analytics Approach to Comparative Network Embedding Inspection
A Survey on Visualization for Explainable Classifier
Understanding Hidden Memories of Recurrent Neural Networks (Yelp Data Challenge Grand Prize Winnner)
CNNComparator: Comparative Analytics of Convolutional Neural Networks
Towards Better Understadning of Deep Learning with Visualization
The Team
Huamin Qu
Professor
Xingbo Wang
Postdoc
Zhihua Jin
Ph.D. Candidate
Yanna Lin
Ph.D. Candidate
Jianben He
Ph.D. Candidate
Rui Sheng
Ph.D. Candidate
Alumni
Furui Cheng
Ph.D. (ETH Zürich Postdoc)
Dong Sun
Ph.D. (NIO)
Yao Ming
Ph.D. (Bloomberg)
Qianwen Wang
Ph.D. (Harvard Postdoc)
Dongyu Liu
Ph.D. (MIT Postdoc)
Haipeng Zeng
Ph.D. (SYSU)
Quan Li
Ph.D. (ShanghaiTech Univ.)
Qiaomu Shen
Ph.D. (Huawei)
Xun Zhao
Ph.D. (Uber)
Collaborators
Address
CYT 3007, Hong Kong University of Science and Technology
Clear Water Bay, Kowloon, Hong Kong
Contact
Zhihua Jin: zjinak at connect dot ust dot hk