doi.bio/davis_liang

Davis Liang

Early Life and Education

Davis Liang is a Staff Research Scientist at Abridge AI (formerly Meta AI and Amazon AI). Liang obtained his MS degree in Computer Science from the University of California, San Diego (UCSD), where he was advised by Prof. Gary Cottrell. During his undergraduate studies, Liang worked with Prof. Michael J Tarr exploring how humans perceive faces and scenes.

Career

Liang is currently working on applying his research in multilingual large language models to healthcare. Previously, he was a Senior Research Scientist at Meta AI, working on natural language processing research. Before that, he worked on information retrieval, machine translation, and speech recognition as an Applied Scientist at Amazon (AWS) AI. Liang has also worked as a Software Engineer at Yahoo.

Research

Liang's research interests include:

• Robust Machine Learning, particularly models that are robust to noises and distribution shifts
• ML for Social Good, tackling issues such as fairness, bias, and misinformation
• Multilingual Systems, with a focus on zero-shot transfer to low-resource languages
• Foundation Models, Generative AI, and unsupervised methods for representation learning

Publications

Liang has numerous publications in the field of machine learning and natural language processing, including:

• "XLM-V: Overcoming the Vocabulary Bottleneck in Multilingual Masked Language Models" (2023)
• "The BELEBELE Benchmark: a Parallel Reading Comprehension Dataset in 122 Language Variants" (2023)
• "Attention-guided generative models for extractive question answering" (2021)
• "Embedding-based Zero-shot Retrieval through Query Generation" (2020)
• "Masked language model scoring" (2020)
• "Improve transformer models with better relative position embeddings" (2020)
• "Learning noise-invariant representations for robust speech recognition" (2018)

- "Deep automated multi-task learning" (2017)

Davis Liang

Early Life and Education

Davis Liang is a researcher and Staff Research Scientist at Abridge AI (formerly Meta AI and Amazon AI). Liang obtained his Master's degree in Computer Science from the University of California, San Diego (UCSD), where he was advised by Prof. Gary Cottrell. During his undergraduate studies, Liang worked with Prof. Michael J Tarr to explore how humans perceive faces and scenes.

Career

Liang has held various roles in the tech industry, including:

• Staff Research Scientist at Abridge AI
• Senior Research Scientist at Meta AI
• Applied Scientist at Amazon AWS AI
• Software Engineer at Yahoo

Research Focus

Liang's research interests include:

• Robust Machine Learning: Developing machine learning models that are robust to noises and distribution shifts.
• ML for Social Good: Tackling issues related to fairness, bias, and misinformation.
• Multilingual Systems: Focusing on zero-shot transfer to low-resource languages.
• Foundation Models, Generative AI, and unsupervised methods for representation learning.

Notable Publications

Liang has numerous publications in the field of AI and machine learning. Some of his notable works include:

• "XLM-V: Overcoming the Vocabulary Bottleneck in Multilingual Masked Language Models" (2023): Liang et al. propose a new approach to scaling large multilingual vocabularies by de-emphasizing token sharing between languages with minimal lexical overlap.
• "The BELEBELE Benchmark: A Parallel Reading Comprehension Dataset in 122 Language Variants" (2023): Bandarkar et al. highlight the benefits of smaller MLMs pretrained on balanced multilingual data for understanding a broader range of languages.
• "Attention-guided Generative Models for Extractive Question Answering" (2021): Xu et al. propose a simple strategy to obtain extractive answer spans from generative models, leveraging decoder cross-attention patterns.
• "Improve Transformer Models with Better Relative Position Embeddings" (2020): Huang et al. propose new methods to encourage increased interaction between query, key, and relative position embeddings in the self-attention mechanism.
• "Learning Noise-Invariant Representations for Robust Speech Recognition" (2018): Liang, Huang, and Lipton introduce invariant-representation-learning (IRL) to improve speech recognition models' robustness against noise.

Youtube Videos

Youtube Title: Davis Liang endorses Kyle Heiskala on Change Politics

Youtube Link: link

Youtube Channel Name: Change Politics

Youtube Channel Link: https://www.youtube.com/@changepolitics4169

Davis Liang endorses Kyle Heiskala on Change Politics

Youtube Title: Bryson Davis vs Liang Ding Zhong Luo Cage Fighting Series 06 on 11.12.2022

Youtube Link: link

Youtube Channel Name: Dragon House MMA

Youtube Channel Link: https://www.youtube.com/@Dragonhousemma

Bryson Davis vs Liang Ding Zhong Luo Cage Fighting Series 06 on 11.12.2022

Youtube Title: Edwaard Liang's The Veil Between Worlds excerpt | Pacific Northwest Ballet

Youtube Link: link

Youtube Channel Name: Pacific Northwest Ballet

Youtube Channel Link: https://www.youtube.com/@pacificnorthwestballet

Edwaard Liang's The Veil Between Worlds excerpt | Pacific Northwest Ballet

Youtube Title: Spark Science from Home Ep. 10: Grow from Plant Clippings

Youtube Link: link

Youtube Channel Name: TELUS Spark

Youtube Channel Link: https://www.youtube.com/@TELUSSparkscience

Spark Science from Home Ep. 10: Grow from Plant Clippings

Youtube Title: Alisha Weir's reaction to learning she got the part of Matilda 🥹 #shorts

Youtube Link: link

Youtube Channel Name: Netflix After School

Youtube Channel Link: https://www.youtube.com/@netflixafterschool

Alisha Weir's reaction to learning she got the part of Matilda 🥹 #shorts

Youtube Title: Spark Science from Home Ep. 8 – What makes us human?

Youtube Link: link

Youtube Channel Name: TELUS Spark

Youtube Channel Link: https://www.youtube.com/@TELUSSparkscience

Spark Science from Home Ep. 8 – What makes us human?

Youtube Title: LIVE from Spark in BODY WORLDS: ANIMAL INSIDE OUT - Plastionation Facts with Davis

Youtube Link: link

Youtube Channel Name: TELUS Spark

Youtube Channel Link: https://www.youtube.com/@TELUSSparkscience

LIVE from Spark in BODY WORLDS: ANIMAL INSIDE OUT - Plastionation Facts with Davis

Youtube Title: Curtis on Tour String Sextet at Mondavi Center UC Davis California 3/3/2024

Youtube Link: link

Youtube Channel Name: Pennia California

Youtube Channel Link: https://www.youtube.com/@penniacalifornia9128

Curtis on Tour String Sextet at Mondavi Center UC Davis California 3/3/2024

Youtube Title: Steve Davis „Interesting" Snooker Shot versus Joe Swail !

Youtube Link: link

Youtube Channel Name: snookasus

Youtube Channel Link: https://www.youtube.com/@snookasus

Steve Davis „Interesting" Snooker Shot versus Joe Swail !

Youtube Title: Finding Janet's Perfect Match at UC Davis

Youtube Link: link

Youtube Channel Name: lowfat422

Youtube Channel Link: https://www.youtube.com/@lowfat422

Finding Janet's Perfect Match at UC Davis

Youtube Title: Final Round — NAMC National Tournament (2023)

Youtube Link: link

Youtube Channel Name: National Association of Moot Court

Youtube Channel Link: https://www.youtube.com/@NAMCMootCourt

Final Round — NAMC National Tournament (2023)

Youtube Title: 15 MOST EMOTIONAL SNOOKER MOMENTS

Youtube Link: link

Youtube Channel Name: SNOOKR

Youtube Channel Link: https://www.youtube.com/@SNOOKR

15 MOST EMOTIONAL SNOOKER MOMENTS

Youtube Title: The Wonder Bucket: What are clouds made of?

Youtube Link: link

Youtube Channel Name: TELUS Spark

Youtube Channel Link: https://www.youtube.com/@TELUSSparkscience

The Wonder Bucket: What are clouds made of?

Youtube Title: Mark King Suspended from World Snooker Tour: What Happened ?

Youtube Link: link

Youtube Channel Name: The Cueist

Youtube Channel Link: https://www.youtube.com/@TheCueist

Mark King Suspended from World Snooker Tour: What Happened ?

Youtube Title: When Steve Davis Faced A 17-Year-Old Ronnie O'Sullivan…

Youtube Link: link

Youtube Channel Name: WST

Youtube Channel Link: https://www.youtube.com/@worldsnookertour

When Steve Davis Faced A 17-Year-Old Ronnie O'Sullivan...

Youtube Title: Moments That Made O'Sullivan & Ding Laugh

Youtube Link: link

Youtube Channel Name: WST

Youtube Channel Link: https://www.youtube.com/@worldsnookertour

Moments That Made O'Sullivan & Ding Laugh

Youtube Title: Times Referees RUINED The Snooker Matches..

Youtube Link: link

Youtube Channel Name: SNOOKR

Youtube Channel Link: https://www.youtube.com/@SNOOKR

Times Referees RUINED The Snooker Matches..