Timothy Liu

NewsMet: A “do it all” Dataset of Contemporary Metaphors in News Headlines

Joseph R., Liu T., Ng A.B., See S., Rai S.

A hand-annotated news headline dataset supports evaluation of how language models interpret and generate metaphors.

Towards Better Characterization of Paraphrases

Liu T. and Soh D.

Two measures describe changes in word order and vocabulary, helping researchers inspect paraphrase datasets and select generated paraphrases.

FedDrip: Federated Learning With Diffusion-Generated Synthetic Image

Huangsuwan K., Liu T., See S., Ng A.B., Vateekul P.

Diffusion-generated images train an additional model to help federated learning cope with differences in data across participating sites.

ESRA: Explainable Scientific Research Assistant

Hongwimol P., Kehasukcharoen P., Laohawarutchai P., Lertvittayakumjorn P., Ng A.B., Lai Z., Liu T., Vateekul P.

A literature search tool uses a knowledge graph to explain returned papers and show connections between research topics.

DAST: A Dual-Stream Voice Anonymization Attacker with Staged Training

Arefeen R., Miao X., Tong R., Ng A.B., See S., Liu T.

The Periodic Table of LLM Reasoning: A Structured Survey of Reasoning Paradigms, Methods, and Failure Modes

Anand A., Ramesh M., Mittal A., Kumar A., Vyalla R.R., Cambria E., Wang Z., Liu T., Ng A.B., See S., Shah R.R.

A structured survey maps LLM reasoning paradigms, methods, and failure modes across more than 300 recent papers.

Advancing Agentic AI: Synthetic Data for Personified and Inclusive Human-AI Interactions

Rajendran M., Tan D., Liu T., Ng A.B., Lee J.S., Wei E.Y., See S.

FORGE: Fused On-Register Gradient Elimination for Memory-Efficient LLM Training

Kukreja D., Prasad K., Anand A., Wang Z., Cambria E., Liu T., Ng A.B., See S., Chatterjee B.

ShalDeepDP: A Survey and Comparison of Data Poisoning in Statistical, Classical Deep Learning and Foundation Models

Chan J.Y., Guo H., Liu T., Ng A.B., See S.

ADAPT: Multibeamformer with Tunable Weight Fusion and Patch-Wise Learning for Ultrasound Imaging

Gopikrishna G., Panicke M.R., Liu T., Ng A.B., See S.

A deep learning multibeamformer uses tunable weight fusion and patch-wise learning for efficient, depth-independent ultrasound reconstruction.