Institution profile

Tufts University

Academic institutionnorthamerica · us
Official website
Research library222linked papers
Opportunities0open roles
Selected work

Representative Papers

How Much Progress Did I Make? An Unexplored Human Feedback Signal for Teaching Robots

Jul 08, 20242024 33rd IEEE International Conference on Robot and Human Interactive Communication (ROMAN)

This work addresses the limitations of conventional human feedback signals—namely, low discriminability, difficulty in quantification, and poor inter-user consistency—for teaching robots to non-expert users. We propose *progress* as a novel, task-agnostic feedback signal, formally defined and empirically validated as a normalized task completion percentage. This signal exhibits strong discriminability between success and failure, enables precise quantification of partial progress, identifies harmless but inefficient behaviors, achieves high cross-user consistency, and imposes zero additional annotation burden. We validate its efficacy and robustness through three empirical studies: an online crowdsourced experiment (N=76), an in-situ user study in public spaces (N=40), and a real-world ice cream topping task. Concurrently, we introduce and publicly release the first benchmark dataset of 40 novice demonstrations, annotated with exploratory actions and operational errors—designed explicitly to support progress-based learning.

3 citationsRead paper

Escape with Your Self: A Solution to the Avoidance Problem with Decidable Bidirectional Typing for Reachability Types

Apr 12, 2024

This work addresses the “escape problem” (variable scope leakage) in reachability-based type systems under subtyping. Existing approaches rely on self-references and function qualifiers, requiring manual term-level coercions—e.g., η-expansions—that impair algorithmic practicality. We propose a refined reachability type system based on self-references, supporting precise static tracking of shared, escaping, and mutable data in higher-order, polymorphic programs under Rust-style ownership semantics. We introduce a novel refined subtyping relation that eliminates all term-level coercions. Furthermore, we present the first sound and decidable bidirectional type-checking algorithm for such systems and formally verify it in Coq. Our results yield a fully decidable, zero-runtime-overhead static reachability inference mechanism, enabling practical lifetime safety guarantees for higher-order languages.

2 citationsRead paper

Advancing the Scientific Method with Large Language Models: From Hypothesis to Discovery

May 22, 2025

Large language models (LLMs) face critical challenges in scientific research—including hallucination, low reliability, and ambiguous ethical accountability—hindering their trustworthy integration into the scientific process. Method: This paper repositions LLMs as “collaborative creative engines” and systematically investigates their deep integration across the full scientific workflow: hypothesis generation → experimental design → data analysis → discovery validation. We synergistically combine prompt engineering, scientific knowledge augmentation, verifiable reasoning-chain construction, and cross-disciplinary workflow integration. Contribution/Results: We introduce (1) the first comprehensive LLM application taxonomy spanning the entire scientific lifecycle; (2) a human-aligned, stage-specific evaluation framework with quantifiable collaboration metrics; and (3) an ethics governance mechanism balancing creative stimulation with responsibility constraints. Our work transcends the conventional instrumental use of LLMs, establishing both theoretical foundations and actionable pathways for AI-augmented scientific paradigm transformation.

1 citationsRead paper

Prediction of Received Power in Low-Power Networks Deployed on the Surface of Rough Waters

Feb 19, 2025

To address link instability and unpredictable received power caused by wave-induced motion in low-power IoT nodes operating over rough water surfaces, this paper proposes a lightweight motion-aware received power prediction model tailored for embedded devices. Methodologically, it replaces the high-complexity matrix inversion in conventional MMSE estimation with online gradient descent for channel parameter estimation; furthermore, it introduces the first joint modeling of surface-node motion statistics—such as pitch and roll distributions—with multipath fading, enabling a time-aware received power prediction framework. Experimental results demonstrate 91% prediction accuracy with minimal iterations, while computational overhead is reduced by two orders of magnitude—enabling real-time, adaptive communication on resource-constrained IoT nodes. Key contributions include: (i) a novel motion-channel joint modeling paradigm; (ii) a low-complexity online parameter estimation mechanism; and (iii) a lightweight prediction architecture specifically designed for dynamic aquatic environments.

1 citationsRead paper

Social AI Agents Too Need to Explain Themselves

Jan 19, 2025International Conference on Intelligent Tutoring Systems

To address the limited explainability of social AI in multi-agent interactions and its consequent difficulty in establishing user trust, this paper proposes a socially situated dynamic explanation generation framework. Methodologically, it pioneers the integration of social cognition theories—including Theory of Mind and Face Theory—into AI explanation mechanisms, synergizing large language models, social relationship graph modeling, and an intent-driven explanation planning module to generate adaptive natural-language explanations conditioned on user roles, relational context, and interaction intent. The framework supports real-time, role-aware explanation delivery and is evaluated via a multi-turn dialogue explainability assessment protocol. On the SocialExplain benchmark, it achieves a 37% improvement in explanation relevance, alongside 29% and 22% gains in user trust and collaborative efficiency, respectively—marking dual advances in social adaptability and interaction consistency.

1 citationsRead paper
Recent publications

Latest Papers

Evidence-Aligned Local Composition of Discrete Experts for Sequence Restoration

Sep 05, 2026

A document modeled as a discrete sequence of tokens can be thought of as being generated from a composition of texts from different domains; a README file, for example, moves between prose, code, and configuration. When such a document is corrupted and only frozen domain experts are available, restoring it requires deciding both what is missing and which expert to trust at each position, at test time and without region labels or a trained router. We introduce evidence-aligned local composition, which infers a soft, position-wise weighting over the experts from the marginal evidence of the corrupted observation under a given corruption model, estimating the evidence from the experts'own denoising losses and smoothing the weights across positions. Because the weighting is soft, it recovers a mixture when the true composition is mixed and concentrates on one expert when that suffices. Across a categorical simulator, byte-level experts, and experts fine-tuned from a $1.3$B discrete flow-matching model, the inferred weights track the true regions at $0.85$ field accuracy on naturally mixed scientific documents, and at $0.98$ on constructed mixtures whose regions are lexically disjoint. Restoration improves over a single global weight when the experts are genuinely distinct and reduces to it when they converge, tracking a measure of expert separation.

0 citationsRead paper