Papers by Shamik Roy

8 papers
Constrained Decoding with Speculative Lookaheads (2025.naacl-long)

Copied to clipboard

Challenge: Constrained decoding with lookahead heuristics is effective for aligning LLM generations to human preferences, but the extensive lookaheaded roll-out operations for each generated token make it prohibitively expensive.
Approach: They propose a technique that uses lookaheads to align LLMs to human preferences . they propose 2.2x to 12.15x speedup over greedy decoding .
Outcome: The proposed technique achieves 2.2x to 12.15x speedup over greedy decoding without significant performance reduction.
Weakly Supervised Learning of Nuanced Frames for Analyzing Polarization in News Media (2020.emnlp-main)

Copied to clipboard

Challenge: a new study suggests a minimally supervised approach for identifying nuanced political frames in news articles on politically divisive topics.
Approach: They propose a minimally supervised approach for identifying nuanced policy frames in news coverage of politically divisive topics.
Outcome: The proposed subframes can capture differences in political ideology better . the proposed frameworks were tested on immigration, gun control and abortion topics .
FLAP: Flow-Adhering Planning with Constrained Decoding in LLMs (2024.naacl-long)

Copied to clipboard

Challenge: Flow-adhering planning algorithm for task oriented dialogs (TODs) is a task-oriented dialog (TO) that can be used for task planning and API usage.
Approach: They propose a Flow-Adhering Planning algorithm that follows predefined flows and preserves API dependencies in task oriented dialogs.
Outcome: The proposed algorithm outperforms other decoding and prompting-based baselines in task oriented dialogs.
Hands-On Interactive Neuro-Symbolic NLP with DRaiL (2022.emnlp-demos)

Copied to clipboard

Challenge: Existing methods to enhance and correct NLP models require feedback from users.
Approach: They propose to enhance DRaiL with an easy to use Python interface that allows users to define, modify and augment models, as well as debug and visualize the predictions.
Outcome: The proposed framework supports predicting sentence and entity level moral sentiment in political tweets.
FairGen: Controlling Sensitive Attributes for Fair Generations in Diffusion Models via Adaptive Latent Guidance (2025.emnlp-main)

Copied to clipboard

Challenge: Text-to-image diffusion models often exhibit generation biases toward specific demographic groups, raising ethical concerns and limiting their adoption.
Approach: They propose an adaptive latent guidance mechanism which controls the generation distribution during inference by dynamically adjusting the diffusion process to enforce specific attributes.
Outcome: The proposed model outperforms existing models on HBE and Stable Bias datasets and achieves substantial bias reduction.
CAIR: Counterfactual-based Agent Influence Ranker for Agentic AI Workflows (2025.emnlp-main)

Copied to clipboard

Challenge: Existing methods to assess the influence of each agent on the AAW’s output perform only static structural analysis, which is unsuitable for inference time execution.
Approach: They propose to use an LLM-based agent influence Ranker to assess the influence level of each agent on the AAW's output and determine which agents are the most influential.
Outcome: The proposed method outperforms baseline methods and produces consistent rankings and relevancy of downstream tasks.
“A Tale of Two Movements’: Identifying and Comparing Perspectives in #BlackLivesMatter and #BlueLivesMatter Movements-related Tweets using Weakly Supervised Graph-based Structured Prediction (2023.findings-emnlp)

Copied to clipboard

Challenge: a weakly supervised graph-based approach to model #BLM-related tweets is difficult to obtain .
Approach: They propose a weakly supervised graph-based approach that explicitly models perspectives in #BackLivesMatter-related tweets.
Outcome: The proposed model outperforms multitask baselines by a large margin.
Identifying Morality Frames in Political Tweets using Relational Learning (2021.emnlp-main)

Copied to clipboard

Challenge: Moral sentiment is often motivated by its targets, which can correspond to individuals or collective entities.
Approach: They propose a model to predict moral attitudes towards entities and moral foundations jointly using tweets written by US politicians.
Outcome: The proposed model predicts moral attitudes towards entities and moral foundations jointly from tweets written by US politicians.

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations