Aneesha Sampath

Machine learning PhD student
focused on multimodal machine learning and affective computing.

I am a second-year PhD student at the Computer Science and Engineering department in the University of Michigan, Ann Arbor. I am advised by Professor Emily Mower Provost.

I received my Bachelor's in Artificial Intelligence from Carnegie Mellon University's School of Computer Science in May 2023. As an undergraduate, I worked on subjectivity uncertainty quantification in emotion recognition, and was advised by Professor Louis-Philippe Morency.

My current research interests include multimodal machine learning and emotion recognition.

Research

I am always interested in collaborating on projects. Please reach out if you would like to collaborate!

Interests

  • Multimodal Machine Learning
  • Emotion Recognition
  • Representation Learning
  • Speech Processing
  • Natural Language Processing
  • Social Intelligence

Current Projects

health
Contextualized Emotion Recognition

Humans are inherently social beings. In emotion prediction, models often isolate the analyzed speech to segments of conversations limited to a single speaker, which disregards the context in which the phrases were communicated. In this project, I am developing techniques to allow models to leverage previously spoken phrases when making emotion predictions.

health
Mental Health Risk Assessment from Natural Phone Conversations

In this project, I am working on developing language and speech models to predict emotion, and ultimately provide an assessment on mental health risk, from natural phone conversations.

Publications

aaai-logo

SeedBERT: Recovering Annotator Rating Distributions from an Aggregated Label [paper]

Aneesha Sampath, Victoria Lin, Louis-Philippe Morency

AAAI-UDM Workshop 2023

Industry Experience

Applied Research Intern, Microsoft Bing (2022)

As a member of the Core Search & AI team, I created two new pipelines for Bing's search result title generation, including a mining-based approach as well as a deep-learning based approach. I also created a new dataset for deep-learning based title generation. I increased the total number of titles by 3% and maintained or increased quality of existing titles in 70% of overriden titles.

Skills: PyTorch, NLP, Deep Learning, SQL

Software Engineer Intern, Microsoft Cloud Monitoring (2021)

As a member of the cloud monitoring team, I designed and developed a querying language from end-to-end to integrate the internal cloud monitoring tools into Jupyter notebooks. The feature reduced context switching and manual query execution throgh automated troubleshooting, and has applicability across most teams at Microsoft that use the troubleshooting service.

Skills: C#, Jupyter

Software Engineer Intern (2020)

As a SWE intern, I developed a full-stack, open-source, multiplayer stock market simulation game from end-to-end as a part of a self-directed team. I coded all aspects of the game's interaction with the database, Cloud Firestore.

Skills: Google Cloud Platform, Cloud Firestore