arXiv subjects

Sailik Sengupta

Publications and source records attributed to Sailik Sengupta.

At least 19 recordsLinked to original sources

arXiv2026

Selective Regenerative Decoding: Trajectory-Level Intervention for Inference-Time Reasoning

Inference-time decoding methods improve LLM reasoning by exploring multiple candidate trajectories, yet treat each trajectory as atomic: either retaining it whole or discarding it irreversibly. This wastes computation on partially promising candidates whose high-quality prefixes are abandoned alongside degraded suffixes. We introduce Selective Regenerative Decoding (SRD), which routes each candidate to discard, keep, or refine only the degraded portion of the suffix while preserving the useful prefix of borderline candidates, without requiring a larger target model. Under mild assumptions, SRD achieves a provable 1.28-to-1.36-fold gain in sample efficiency over rejection sampling with strictly higher expected trajectory quality, with the gain growing as the candidate pool grows. Across MATH500, GPQA Diamond, HotpotQA, and AlpacaEval with multiple generation-reward model pairs, SRD matches Best-of-N accuracy with substantially fewer generated tokens and outperforms speculative rejection in low-compute regimes. By enabling segment-level intervention rather than whole-trajectory selection, SRD opens a previously underexplored region of the accuracy-compute tradeoff for inference-time reasoning.

cs.AI↗

arXiv2026

{\epsilon}-Indistinguishability In Moving Target Defense: Framework, Algorithms, And Cloud Case Studies

Moving Target Defense (MTD) assumes its pool of candidate configurations is safe to cycle among, i.e. latency and other observables do not trivially fingerprint the active choice, but this assumption has not been quantified at the pool level. We formalize this pool-safety problem as finding the largest $\varepsilon$-close subset of the Cartesian product of per-component implementation choices, reducing pairwise indistinguishability under an additive utility model to a densest-window query over a sum-set. We give four algorithms spanning the scalability spectrum -- full enumeration, meet-in-the-middle, FFT convolution, and Monte Carlo sampling -- covering configuration spaces from tens to $10^{38}$. We then measure the anonymity gap end-to-end on two production cloud case studies, and find that a component's latency differences do not survive deployment unchanged: a four-runtime serverless rotation, where nothing else masks the interpreter, collapses from four-way to three-way anonymity against a VPC-adjacent adversary, while a $27$-configuration three-tier stack, where the same interpreter differences are instead absorbed by a shared $8$~ms database round-trip, delivers nine-way effective anonymity. The framework and the two case studies together suggest a diagnostic for MTD design: rotating a component adds anonymity only if the latency differences among its variants are too small for the adversary to identify.

cs.CR↗

arXiv2025

InsTALL: Context-aware Instructional Task Assistance with Multi-modal Large Language Models

The improved competence of generative models can help building multi-modal virtual assistants that leverage modalities beyond language. By observing humans performing multi-step tasks, one can build assistants that have situational awareness of actions and tasks being performed, enabling them to cater assistance based on this understanding. In this paper, we develop a Context-aware Instructional Task Assistant with Multi-modal Large Language Models (InsTALL) that leverages an online visual stream (e.g. a user's screen share or video recording) and responds in real-time to user queries related to the task at hand. To enable useful assistance, InsTALL 1) trains a multi-modal model on task videos and paired textual data, and 2) automatically extracts task graph from video data and leverages it at training and inference time. We show InsTALL achieves state-of-the-art performance across proposed sub-tasks considered for multimodal activity understanding -- task recognition (TR), action recognition (AR), next action prediction (AP), and plan prediction (PP) -- and outperforms existing baselines on two novel sub-tasks related to automatic error identification.

cs.CV↗

arXiv2024

Multi-Objective Alignment of Large Language Models Through Hypervolume Maximization

Multi-objective alignment from human feedback (MOAHF) in large language models (LLMs) is a challenging problem as human preferences are complex, multifaceted, and often conflicting. Recent works on MOAHF considered a-priori multi-objective optimization (MOO), where human preferences are known at training or inference time. In contrast, when human preferences are unknown or difficult to quantify, a natural approach is to cover the Pareto front by multiple diverse solutions. We propose an algorithm HaM for learning diverse LLM policies that maximizes their hypervolume. This is the first application of a-posteriori MOO to MOAHF. HaM is computationally and space efficient, and empirically superior across objectives such as harmlessness, helpfulness, humor, faithfulness, and hallucination, on various datasets.

cs.LG↗

arXiv2024

SeRA: Self-Reviewing and Alignment of Large Language Models using Implicit Reward Margins

Direct alignment algorithms (DAAs), such as direct preference optimization (DPO), have become popular alternatives for Reinforcement Learning from Human Feedback (RLHF) due to their simplicity, efficiency, and stability. However, the preferences used in DAAs are usually collected before the alignment training begins and remain unchanged (off-policy). This can lead to two problems where the policy model (1) picks up on spurious correlations in the dataset (as opposed to learning the intended alignment expressed in the human preference labels), and (2) overfits to feedback on off-policy trajectories that have less likelihood of being generated by an updated policy model. To address these issues, we introduce Self-Reviewing and Alignment (SeRA), a cost-efficient and effective method that can be readily combined with existing DAAs. SeRA comprises of two components: (1) sample selection using implicit reward margins, which helps alleviate over-fitting to some undesired features, and (2) preference bootstrapping using implicit rewards to augment preference data with updated policy models in a cost-efficient manner. Extensive experimentation, including some on instruction-following tasks, demonstrate the effectiveness and generality of SeRA in training LLMs on offline preference datasets with DAAs.

cs.LG↗

arXiv2024

FLAP: Flow-Adhering Planning with Constrained Decoding in LLMs

Planning is a crucial task for agents in task oriented dialogs (TODs). Human agents typically resolve user issues by following predefined workflows, decomposing workflow steps into actionable items, and performing actions by executing APIs in order; all of which require reasoning and planning. With the recent advances in LLMs, there have been increasing attempts to use them for task planning and API usage. However, the faithfulness of the plans to predefined workflows and API dependencies, is not guaranteed with LLMs. Moreover, workflows in real life are often custom-defined and prone to changes; hence, adaptation is desirable. To study this, we propose the problem of faithful planning in TODs that needs to resolve user intents by following predefined flows and preserving API dependencies. To solve this problem, we propose FLAP, a Flow-Adhering Planning algorithm based on constrained decoding with lookahead heuristic for LLMs. Our algorithm alleviates the need for finetuning LLMs using domain specific (plan/dependency) data, enables quick adaptation to predefined flows, and outperforms other decoding and prompting-based baselines. Further, our algorithm empowers smaller LLMs (7B) to perform at par larger LLMs (30B-40B).

cs.CL↗

arXiv2024

Can Your Model Tell a Negation from an Implicature? Unravelling Challenges With Intent Encoders

Conversational systems often rely on embedding models for intent classification and intent clustering tasks. The advent of Large Language Models (LLMs), which enable instructional embeddings allowing one to adjust semantics over the embedding space using prompts, are being viewed as a panacea for these downstream conversational tasks. However, traditional evaluation benchmarks rely solely on task metrics that don't particularly measure gaps related to semantic understanding. Thus, we propose an intent semantic toolkit that gives a more holistic view of intent embedding models by considering three tasks -- (1) intent classification, (2) intent clustering, and (3) a novel triplet task. The triplet task gauges the model's understanding of two semantic concepts paramount in real-world conversational systems -- negation and implicature. We observe that current embedding models fare poorly in semantic understanding of these concepts. To address this, we propose a pre-training approach to improve the embedding model by leveraging augmentation with data generated by an auto-regressive model and a contrastive loss term. Our approach improves the semantic understanding of the intent embedding model on the aforementioned linguistic dimensions while slightly effecting their performance on downstream task metrics.

cs.CL↗

arXiv2024

DeAL: Decoding-time Alignment for Large Language Models

Large Language Models (LLMs) are nowadays expected to generate content aligned with human preferences. Current work focuses on alignment at model training time, through techniques such as Reinforcement Learning with Human Feedback (RLHF). However, it is unclear if such methods are an effective choice to teach alignment objectives to the model. First, the inability to incorporate multiple, custom rewards and reliance on a model developer's view of universal and static principles are key limitations. Second, the reliability of such approaches is also questionable (e.g. susceptibility to jailbreaking even after safety training). To address these issues, we propose DeAL, a framework that allows the user to customize reward functions and enables Decoding-time Alignment of LLMs (DeAL). At its core, we view decoding as a heuristic-guided search process and facilitate the use of a wide variety of alignment objectives. Our experiments with programmatic constraints such as keyword and length constraints, and abstract objectives such as harmlessness and helpfulness, show that we can DeAL with fine-grained trade-offs and improve adherence to alignment objectives. Lastly, we demonstrate that DeAL is largely complementary to existing alignment strategies, and can be effectively paired with RLHF and prompting techniques to achieve better alignment.

cs.AI↗

arXiv2023

`Why didn't you allocate this task to them?' Negotiation-Aware Explicable Task Allocation and Contrastive Explanation Generation

Task allocation is an important problem in multi-agent systems. It becomes more challenging when the team-members are humans with imperfect knowledge about their teammates' costs and the overall performance metric. In this paper, we propose a centralized Artificial Intelligence Task Allocation (AITA) that simulates a negotiation and produces a negotiation-aware explicable task allocation. If a team-member is unhappy with the proposed allocation, we allow them to question the proposed allocation using a counterfactual. By using parts of the simulated negotiation, we are able to provide contrastive explanations that provide minimum information about other's cost to refute their foil. With human studies, we show that (1) the allocation proposed using our method appears fair to the majority, and (2) when a counterfactual is raised, explanations generated are easy to comprehend and convincing. Finally, we empirically study the effect of different kinds of incompleteness on the explanation-length and find that underestimation of a teammate's costs often increases it.

cs.AI↗

arXiv2023

Measuring and Mitigating Constraint Violations of In-Context Learning for Utterance-to-API Semantic Parsing

In executable task-oriented semantic parsing, the system aims to translate users' utterances in natural language to machine-interpretable programs (API calls) that can be executed according to pre-defined API specifications. With the popularity of Large Language Models (LLMs), in-context learning offers a strong baseline for such scenarios, especially in data-limited regimes. However, LLMs are known to hallucinate and therefore pose a formidable challenge in constraining generated content. Thus, it remains uncertain if LLMs can effectively perform task-oriented utterance-to-API generation where respecting API's structural and task-specific constraints is crucial. In this work, we seek to measure, analyze and mitigate such constraints violations. First, we identify the categories of various constraints in obtaining API-semantics from task-oriented utterances, and define fine-grained metrics that complement traditional ones. Second, we leverage these metrics to conduct a detailed error analysis of constraints violations seen in state-of-the-art LLMs, which motivates us to investigate two mitigation strategies: Semantic-Retrieval of Demonstrations (SRD) and API-aware Constrained Decoding (API-CD). Our experiments show that these strategies are effective at reducing constraints violations and improving the quality of the generated API calls, but require careful consideration given their implementation complexity and latency.

cs.AI↗

arXiv2023

Robustification of Multilingual Language Models to Real-world Noise in Crosslingual Zero-shot Settings with Robust Contrastive Pretraining

Advances in neural modeling have achieved state-of-the-art (SOTA) results on public natural language processing (NLP) benchmarks, at times surpassing human performance. However, there is a gap between public benchmarks and real-world applications where noise, such as typographical or grammatical mistakes, is abundant and can result in degraded performance. Unfortunately, works which evaluate the robustness of neural models on noisy data and propose improvements, are limited to the English language. Upon analyzing noise in different languages, we observe that noise types vary greatly across languages. Thus, existing investigations do not generalize trivially to multilingual settings. To benchmark the performance of pretrained multilingual language models, we construct noisy datasets covering five languages and four NLP tasks and observe a clear gap in the performance between clean and noisy data in the zero-shot cross-lingual setting. After investigating several ways to boost the robustness of multilingual models in this setting, we propose Robust Contrastive Pretraining (RCP). RCP combines data augmentation with a contrastive loss term at the pretraining stage and achieves large improvements on noisy (and original test data) across two sentence-level (+3.2%) and two sequence-labeling (+10 F1-score) multilingual classification tasks.

cs.CL↗

arXiv2022

Parameter and Data Efficient Continual Pre-training for Robustness to Dialectal Variance in Arabic

The use of multilingual language models for tasks in low and high-resource languages has been a success story in deep learning. In recent times, Arabic has been receiving widespread attention on account of its dialectal variance. While prior research studies have tried to adapt these multilingual models for dialectal variants of Arabic, it still remains a challenging problem owing to the lack of sufficient monolingual dialectal data and parallel translation data of such dialectal variants. It remains an open problem on whether the limited dialectical data can be used to improve the models trained in Arabic on its dialectal variants. First, we show that multilingual-BERT (mBERT) incrementally pretrained on Arabic monolingual data takes less training time and yields comparable accuracy when compared to our custom monolingual Arabic model and beat existing models (by an avg metric of +$6.41$). We then explore two continual pre-training methods -- (1) using small amounts of dialectical data for continual finetuning and (2) parallel Arabic to English data and a Translation Language Modeling loss function. We show that both approaches help improve performance on dialectal classification tasks ($+4.64$ avg. gain) when used on monolingual models.

cs.CL↗

arXiv2022

RADAR-X: An Interactive Mixed Initiative Planning Interface Pairing Contrastive Explanations and Revised Plan Suggestions

Decision support systems seek to enable informed decision-making. In the recent years, automated planning techniques have been leveraged to empower such systems to better aid the human-in-the-loop. The central idea for such decision support systems is to augment the capabilities of the human-in-the-loop with automated planning techniques and enhance the quality of decision-making. In addition to providing planning support, effective decision support systems must be able to provide intuitive explanations based on specific user queries for proposed decisions to its end users. Using this as motivation, we present our decision support system RADAR-X that showcases the ability to engage the user in an interactive explanatory dialogue by first enabling them to specify an alternative to a proposed decision (which we refer to as foils), and then providing contrastive explanations to these user-specified foils which helps the user understand why a specific plan was chosen over the alternative (or foil). Furthermore, the system uses this dialogue to elicit the user's latent preferences and provides revised plan suggestions through three different interaction strategies.

cs.AI↗

arXiv2022

Inference of Human's Observation Strategy for Monitoring Robot's Behavior based on a Game-Theoretic Model of Trust

We consider scenarios where a worker robot, who may be unaware of the human's exact expectations, may have the incentive to deviate from a preferred plan (e.g. safe but costly) when a human supervisor is not monitoring it. On the other hand, continuous monitoring of the robot's behavior is often difficult for humans because it costs them valuable resources (e.g., time, cognitive overload, etc.). Thus, to optimize the cost of monitoring while ensuring the robots follow the {\em safe} behavior and to assist the human to deal with the possible unsafe robots, we model this problem in a game-theoretic framework of trust. In settings where the human does not initially trust the robot, pure-strategy Nash Equilibrium provides a useful policy for the human. Unfortunately, we show the formulated game often lacks a pure strategy Nash equilibrium. Thus, we define the concept of a trust boundary over the mixed strategy space of the human and show that it helps to discover optimal monitoring strategies. We conduct humans subject studies that demonstrate (1) the need for coming up with optimal monitoring strategies, and (2) the benefits of using strategies suggested by our approach.

cs.AI↗

arXiv2021

Not all Failure Modes are Created Equal: Training Deep Neural Networks for Explicable (Mis)Classification

Deep Neural Networks are often brittle on image classification tasks and known to misclassify inputs. While these misclassifications may be inevitable, all failure modes cannot be considered equal. Certain misclassifications (eg. classifying the image of a dog to an airplane) can perplex humans and result in the loss of human trust in the system. Even worse, these errors (eg. a person misclassified as a primate) can have odious societal impacts. Thus, in this work, we aim to reduce inexplicable errors. To address this challenge, we first discuss methods to obtain the class-level semantics that capture the human's expectation ($M^h$) regarding which classes are semantically close {\em vs.} ones that are far away. We show that for popular image benchmarks (like CIFAR-10, CIFAR-100, ImageNet), class-level semantics can be readily obtained by leveraging either human subject studies or publicly available human-curated knowledge bases. Second, we propose the use of Weighted Loss Functions (WLFs) to penalize misclassifications by the weight of their inexplicability. Finally, we show that training (or fine-tuning) existing classifiers with the proposed methods lead to Deep Neural Networks that have (1) comparable top-1 accuracy, (2) more explicable failure modes on both in-distribution and out-of-distribution (OOD) test data, and (3) incur significantly less cost in the gathering of additional human labels compared to existing works.

cs.LG↗

arXiv2021

On the Robustness of Intent Classification and Slot Labeling in Goal-oriented Dialog Systems to Real-world Noise

Intent Classification (IC) and Slot Labeling (SL) models, which form the basis of dialogue systems, often encounter noisy data in real-word environments. In this work, we investigate how robust IC/SL models are to noisy data. We collect and publicly release a test-suite for seven common noise types found in production human-to-bot conversations (abbreviations, casing, misspellings, morphological variants, paraphrases, punctuation and synonyms). On this test-suite, we show that common noise types substantially degrade the IC accuracy and SL F1 performance of state-of-the-art BERT-based IC/SL models. By leveraging cross-noise robustness transfer -- training on one noise type to improve robustness on another noise type -- we design aggregate data-augmentation approaches that increase the model performance across all seven noise types by +10.8% for IC accuracy and +15 points for SL F1 on average. To the best of our knowledge, this is the first work to present a single IC/SL model that is robust to a wide range of noise phenomena.

cs.CL↗

arXiv2021

Imperfect ImaGANation: Implications of GANs Exacerbating Biases on Facial Data Augmentation and Snapchat Selfie Lenses

In this paper, we show that popular Generative Adversarial Networks (GANs) exacerbate biases along the axes of gender and skin tone when given a skewed distribution of face-shots. While practitioners celebrate synthetic data generation using GANs as an economical way to augment data for training data-hungry machine learning models, it is unclear whether they recognize the perils of such techniques when applied to real world datasets biased along latent dimensions. Specifically, we show that (1) traditional GANs further skew the distribution of a dataset consisting of engineering faculty headshots, generating minority modes less often and of worse quality and (2) image-to-image translation (conditional) GANs also exacerbate biases by lightening skin color of non-white faces and transforming female facial features to be masculine when generating faces of engineering professors. Thus, our study is meant to serve as a cautionary tale.

cs.LG↗

arXiv2020

Moving Target Defense for Robust Monitoring of Electric Grid Transformers in Adversarial Environments

Electric power grid components, such as high voltage transformers (HVTs), generating stations, substations, etc. are expensive to maintain and, in the event of failure, replace. Thus, regularly monitoring the behavior of such components is of utmost importance. Furthermore, the recent increase in the number of cyberattacks on such systems demands that such monitoring strategies should be robust. In this paper, we draw inspiration from work in Moving Target Defense (MTD) and consider a dynamic monitoring strategy that makes it difficult for an attacker to prevent unique identification of behavioral signals that indicate the status of HVTs. We first formulate the problem of finding a differentially immune configuration set for an MTD in the context of power grids and then propose algorithms to compute it. To find the optimal movement strategy, we model the MTD as a two-player game and consider the Stackelberg strategy. With the help of IEEE test cases, we show the efficacy and scalability of our proposed approaches.

cs.GT↗