Ethan Liu

Blind and Sight Impaired Society

Author Statistics

Papers

Citation

H-Index

i-10 index

Research Trends

Author Order

Document Type

Co-Authors

C. Kevin S. Zhou

Institute of Computing Technology

Nesreen K. Ahmed

Intel (United States)

Franck Dernoncourt

Adobe Systems (United States)

Junhua Fang

First People's Hospital of Yunnan Province

B. Yifan Chen

Guangdong University of Foreign Studies

F. Thomas S. Huang

Huaqiao University

Nedim Lipka

Adobe Systems (United States)

D. Bogdan Georgescu

Siemens Healthcare (United States)

Jieun Kim

Hanyang University

Nikhil Singh

Massachusetts Institute of Technology

Cooperative Institutions

Adobe Systems (United States)

University of Maryland, College Park

Purdue University West Lafayette

Weatherford College

Paderborn University

Key Laboratory of Nuclear Radiation and Nuclear Energy Technology

University of Illinois Urbana-Champaign

Northwestern University

Southern California University for Professional Studies

University of Southern California

Author Statistics

Papers

Citation

H-Index

i-10 index

Research Field

Multi-LLM Text Summarization

arXiv (Cornell University) (2024)

Junhua Fang C.-L. Liu Jieun Kim Yash Bhedaru Ethan Liu

In this work, we propose a Multi-LLM summarization framework, and investigate two different multi-LLM strategies including centralized and decentralized. Our multi-LLM summarization framework has two fundamentally important steps at each round of conversation: generation and evaluation. These steps are different depending on whether our multi-LLM decentralized summarization is used or centralized. In both our multi-LLM decentralized and centralized strategies, we have k different LLMs that generate diverse summaries of the text. However, during evaluation, our multi-LLM centralized summarization approach leverages a single LLM to evaluate the summaries and select the best one whereas k LLMs are used for decentralized multi-LLM summarization. Overall, we find that our multi-LLM summarization approaches significantly outperform the baselines that leverage only a single LLM by up to 3x. These results indicate the effectiveness of multi-LLM approaches for summarization.

10.48550/arxiv.2412.15487

Cite

Citations (0)

Learning A Self-Inverse Network for Bidirectional Mri Image Synthesis

2022 IEEE 19th International Symposium on Biomedical Imaging (ISBI) (2020)

Ao Shen B. Yifan Chen C. Kevin S. Zhou D. Bogdan Georgescu Ethan Liu

The one-to-one mapping is necessary for MRI image synthesis as MRI images are unique to the patient. State-of-the-art approaches for image synthesis from domain X to domain Y learn a convolutional neural network that meticulously maps between the domains. A different network is typically implemented to map along the opposite direction, from Y to X. In this paper, we explore the possibility of only wielding one network for bi-directional image synthesis. In other words, such an autonomous learning network implements a self-inverse function. A self-inverse network shares several distinct advantages: only one network instead of two, better generalization and more restricted parameter space. Most importantly, a self-inverse function guarantees a one-to-one mapping, a property that cannot be guaranteed by earlier approaches that are not self-inverse. The experiments on MRI T1 and T2 images show that, compared with the baseline approaches that use two separate models for the image synthesis along two directions, our self-inverse network achieves better synthesis results in terms of standard metrics. Finally, our sensitivity analysis confirms the feasibility of learning a one-to-one mapping function for MRI image synthesis.

Inverse function

10.1109/isbi45749.2020.9098576

Cite

Citations (1)

Simplifying Multimodal Composition: A Novel Zero-shot Framework to Visual Question Answering and Image Captioning

Research Square (Research Square) (2023)

Ethan Liu

Abstract In this work, we perform zero-shot image captioning and visual question answering on images using a simple model composition framework, composing the dense image captioning capabilities of a visual language model with the powerful reasoning abilities of a large language model, ChatGPT. The proposed method utilizes zero-shot learning to enable cross-modal integration of vision and language in order to create a comprehensive visual language model. We achieve zero-shot state-of-the-art performance on VQAv2, demonstrating its effectiveness and high accuracy. The method's simplicity makes it highly scalable and adaptable to a wide range of applications, including integration from OpenAI’s multimodal model, GPT-4, with audio language models in the future. The results demonstrate the vast potential of this simple zero-shot framework in improving the accuracy and relevance of vision and language applications, constituting an effective approach to image captioning and visual question answering, as well as future multimodal composition.

Closed captioning

Simplicity

Relevance

Zero (linguistics)

10.21203/rs.3.rs-3027308/v1

Cite

Citations (0)