PUBLICATION

Transductive Reasoning : resolving aleatoric uncertainty through information tree search during test time generation
July 2025
Effcient Mathematical Reasoning : When SLM cannot solve, but LLMs do. uncertainty-based teacher call
July 2025
Three Level Approach : how to interpret the intermediate state when it comes to subject concepts in formality transfer task.
July 2025
LLMs are fundamentally vulnerable to jailbreak attack. How to implement and how to mitigate ?
July 2025
Story Generation : how to make continual and coherent visual stories but render users immersive.
February 2025
Differentially Privacy : prevent data hacking from database by cluster-aggregation. Keep client’s information safe.
February 2025
Diffusion-EAGS : what is the advantage of conditional [MASK] language model compared to GPT-based LLMs ?
February 2025
Inductive reasoning or Memorial Reasoning ? making diverse hypothesis and select best fit hypothesis to solve programming synthesis.
November 2024
Visual Large Language Models : They are dependent on language prior knowledge which is acquired in pre-training stage.
November 2024
Reliable Agent : intent induction and toxicity mitigation for reliable LM agent
June 2024
LATTE : can LLMs recognize the concept of toxicity ? Definition-based toxicity metric
June 2024
LLMs’ Gender Fairness : fine-grained gender control in machine translation with large language models
Januray 2024
Programming By Example : weakly supervised semantic parsing with execution-based spurious program filtering
May 2023
Speech Synthesis with Prosody : diversifying prosodic features of speech via determinantal point process
May 2023
Encoder Tuning is Enough : target-agnostic gender-aware contrastive learning for mitigating bias in multilingual machine translation
May 2023
DSTC11 : leveraging ensemble techniques and metadata for subjective knowledge grounded conversational systems
December 2022
DSTC11 : multi-view zero-shot open intent induction from dialogues, multi-domain batch and proxy gradient transfer
December 2022

PUBLICATION

Transductive Reasoning : resolving aleatoric uncertainty through information tree search during test time generation

Effcient Mathematical Reasoning : When SLM cannot solve, but LLMs do. uncertainty-based teacher call

Three Level Approach : how to interpret the intermediate state when it comes to subject concepts in formality transfer task.

LLMs are fundamentally vulnerable to jailbreak attack. How to implement and how to mitigate ?

Story Generation : how to make continual and coherent visual stories but render users immersive.

Differentially Privacy : prevent data hacking from database by cluster-aggregation. Keep client’s information safe.

Diffusion-EAGS : what is the advantage of conditional [MASK] language model compared to GPT-based LLMs ?

Inductive reasoning or Memorial Reasoning ? making diverse hypothesis and select best fit hypothesis to solve programming synthesis.

Visual Large Language Models : They are dependent on language prior knowledge which is acquired in pre-training stage.

Reliable Agent : intent induction and toxicity mitigation for reliable LM agent

LATTE : can LLMs recognize the concept of toxicity ? Definition-based toxicity metric

LLMs’ Gender Fairness : fine-grained gender control in machine translation with large language models

Programming By Example : weakly supervised semantic parsing with execution-based spurious program filtering

Speech Synthesis with Prosody : diversifying prosodic features of speech via determinantal point process

Encoder Tuning is Enough : target-agnostic gender-aware contrastive learning for mitigating bias in multilingual machine translation

DSTC11 : leveraging ensemble techniques and metadata for subjective knowledge grounded conversational systems

DSTC11 : multi-view zero-shot open intent induction from dialogues, multi-domain batch and proxy gradient transfer