About Me

• I am a Singapore Citizen.

• I am currently working as a Senior Manager in Data Science at American Express (Amex), Singapore, focusing on Generative AI, Large Language Models (LLMs), and NLP applications in financial risk management.

• Previously, I worked as a Senior Scientist at the Institute for Infocomm Research (I²R), Agency for Science, Technology and Research (A*STAR), Singapore, conducting applied AI research in computer vision, deep learning, and multimodal learning.

• I obtained my Bachelor’s degree from Beijing Institute of Technology in 2013 and my Ph.D. from Nanyang Technological University (NTU), Singapore, in 2019. I joined A*STAR as a Research Scientist in 2018 and spent over seven years there before joining American Express.

• My current research interests include Generative AI, Large Language Models, Vision-Language Models, Multimodal Learning, and Computer Vision, with a focus on both theoretical advances and real-world applications.

Recent News

• One paper entitled “PEVA-Net: Prompt-enhanced view aggregation network for zero/few-shot multi-view 3D shape recognition” has been published in Neurocomputing.

• One paper entitled “PartCLIP: How does CLIP assist mechanical part image retrieval?” has been accepted in ICME2024.

• My Research Work on Multi-View 3D Object Recognition and Retrieval is shown in A*STAR Research Highlight Page Link

• My Research Work on Defect Detection and Segmentation is shown in A*STAR Research Highlight Page Link

• One paper entitled “SCA-PVNet: Self-and-cross attention based aggregation of point cloud and multi-view for 3D object retrieval” has been published in Knowledge-Based Systems.

• One paper entitled “Keyword-Aware Relative Spatio-Temporal Graph Networks for Video Question Answering” has been published in IEEE Transactions on Multimedia.

• One US Patent entitled “Method and system for image classification” has been granted with Patent Number: 11836632.

• Our solution has won the 1st Place in EPIC-Kitchens Dataset Challenges: Unsupervised Domain Adaptation for Action Recognition Track, CVPR2023 Leaderboard Link

• One paper entitled “DDR-ID: Dual deep reconstruction networks based image decomposition for anomaly detection” has been published in Journal of Ambient Intelligence and Humanized Computing.

• One paper entitled “Multi-range view aggregation network with vision transformer feature fusion for 3D object retrieval” has been published in IEEE Transactions on Multimedia.

• One paper entitled “An effective industrial defect classification method under the few-shot setting via two-stream training” has been published in Optics and Lasers in Engineering.

• One paper entitled “Biomedical image classification based on a feature concatenation and ensemble of deep CNNs” has been published in Journal of Ambient Intelligence and Humanized Computing.

• Three papers entitled “MLSA-UNet: End-to-End multi-level spatial attention guided UNet for industrial defect segmentation”, “Masked face recognition via self-attention based local consistency regularization” and “Implicit shape biased few-shot learning for 3d object generalization” have been published in ICIP2022.

• One paper entitled “Multi-view 3D object retrieval leveraging the aggregation of view and instance attentive features” has been published in Knowledge-Based Systems.

• One paper entitled “On the use of component structural characteristics for voxel segmentation in semicon 3D images” has been published in ICASSP2022.

• Our solution has won the 1st Place in EPIC-Kitchens Dataset Challenges: Unsupervised Domain Adaptation for Action Recognition Track, CVPR2022 Leaderboard Link

• One paper entitled “A progressive multi-view learning approach for multi-loss optimization in 3d object recognition” has been published in IEEE Signal Processing Letters.

• One paper entitled “CAM-guided Multi-Path Decoding U-Net with Triplet Feature Regularization for defect detection and segmentation” has been published in Knowledge-Based Systems.

• Two papers entitled “CAM-Guided u-net with adversarial regularization for defect segmentation” and “Action Relational Graph for Weakly-Supervised Temporal Action Localization” have been published in ICIP2021.

• One paper entitled “Few-shot defect segmentation leveraging abundant defect-free training samples through normal background regularization and crop-and-paste operation” has been published in ICME2021(Oral Presentation).

• One paper entitled “Rethinking of deep models parameters with respect to data distribution” has been published in ICPR2021.

• One paper entitled “maskedfacenet: A progressive semi-supervised masked face detector” has been published in WACV2021.

• Two papers entitled “CAM-UNET: class activation MAP guided UNET with feedback refinement for defect segmentation” and “Improving 3d brain tumor segmentation with predict-refine mechanism using saliency and feature maps” have been published in ICIP2020.

• One paper entitled “RefineU-Net: Improved U-Net with progressive global feedbacks and residual attention guided local refinement for medical image segmentation” has been published in Pattern Recognition Letters.

• One paper entitled “Discriminative features for incremental learning classifier” has been published in ICIP2019.

• One paper entitle “A two-stage method for automated detection of ring-like endosomes in fluorescent microscopy images” has been published in Plos One.

• One paper entitled “Vibration source classification and propagation distance estimation system based on spectrogram and KELM” has been published in Cognitive Computation and Systems.

• One paper entitled “Meta module generation for fast few-shot incremental learning” has been published in ICCVW2019.

• One paper entitled “Multicomponent signal decomposition using morphological operations” has been published in DSP2018.

• One paper entitled “Deep CNNs for microscopic image classification by exploiting transfer learning and feature concatenation” has been published in ISCAS2018.

• One paper entitled “Biomedical image classification based on a cascade of an SVM with a reject option and subspace analysis” has been published in Computers in biology and medicine.

• One paper entitled “Twin SVM with a reject option through ROC curve” has been published in Journal of the Franklin Institute.

• Two papers entitled “Automatic endosomal structure detection and localization in fluorescence microscopic images” and “LLC encoded BoW features and softmax regression for microscopic image classification” have been published in ISCAS2017.

• One paper entitled “Reaction-diffusion based level set method with local entropy thresholding for melasma image segmentation” has been published in ISCAS2017.

• One paper entiteld “An SVM based scoring evaluation system for fluorescence microscopic image classification” has been published in DSP2015.