arXiv subjects

Shashank Srivastava

Publications and source records attributed to Shashank Srivastava.

At least 37 records · Page 2Linked to original sources

arXiv2025

Algorithmic Improvements to List Decoding of Folded Reed-Solomon Codes

Folded Reed-Solomon (FRS) codes are a well-studied family of codes, known for achieving list decoding capacity. In this work, we give improved deterministic and randomized algorithms for list decoding FRS codes of rate $R$ up to radius $1-R-\varepsilon$. We present a deterministic decoder that runs in near-linear time $\widetilde{O}_{\varepsilon}(n)$, improving upon the best-known runtime $n^{Ω(1/\varepsilon)}$ for decoding FRS codes. Prior to our work, no capacity achieving code was known whose deterministic decoding could be done in time $\widetilde{O}_{\varepsilon}(n)$. We also present a randomized decoder that runs in fully polynomial time $\mathrm{poly}(1/\varepsilon) \cdot \widetilde{O}(n)$, improving the best-known runtime $\mathrm{exp}(1/\varepsilon)\cdot \widetilde{O}(n)$ for decoding FRS codes. Again, prior to our work, no capacity achieving code was known whose decoding time depended polynomially on $1/\varepsilon$. Our results are based on improved pruning procedures for finding the list of codewords inside a constant-dimensional affine subspace.

cs.IT↗

arXiv2025

List Decoding Expander-Based Codes up to Capacity in Near-Linear Time

We give a new framework based on graph regularity lemmas, for list decoding and list recovery of codes based on spectral expanders. Using existing algorithms for computing regularity decompositions of sparse graphs in (randomized) near-linear time, and appropriate choices for the constant-sized inner/base codes, we prove the following: - Expander-based codes constructed using the distance amplification technique of Alon, Edmonds and Luby [FOCS 1995] with rate $ρ$, can be list decoded to a radius $1 - ρ- ε$ in near-linear time. By known results, the output list has size $O(1/ε)$. - The above codes of Alon, Edmonds and Luby, with rate $ρ$, can also be list recovered to radius $1 - ρ- ε$ in near-linear time, with constant-sized output lists. - The Tanner code construction of Sipser and Spielman [IEEE Trans. Inf. Theory 1996] with distance $δ$, can be list decoded to radius $δ- ε$ in near-linear time, with constant-sized output lists. Our results imply novel combinatorial as well as algorithmic bounds for each of the above explicit constructions. All of these bounds are obtained via combinatorial rigidity phenomena, proved using (weak) graph regularity. The regularity framework allows us to lift the list decoding and list recovery properties for the local base codes, to the global codes obtained via the above constructions.

cs.DS↗

arXiv2025

Ising superconductivity in bulk layered non-centrosymmetric 4H-NbSe2

Transition metal dichalcogenides exhibit multiple polymorphs that enable the exploration of diverse quantum states, including valley-selective spin polarization, the valley Hall effect, Ising superconductivity, and nontrivial topology. Monolayer 2$H$-NbSe$_2$ is a promising candidate for realizing Ising superconductivity due to its spin-split, out-of-plane spin-polarized states arising from inversion symmetry breaking and strong spin-orbit coupling. In contrast, bulk 2$H$-NbSe$_2$ retains inversion symmetry and lacks spin splitting, limiting its suitability for hosting Ising superconductivity. Here, we report the growth of high-quality single crystals of the acentric bulk superconducting polymorph, 4$H$-NbSe$_2$, which intrinsically breaks the inversion symmetry and supports valley-selective spin-polarized states. Magnetization and resistivity measurements reveal anisotropic superconductivity, with the in-plane upper critical field exceeding the Pauli limit, while out-of-plane fields suppress superconductivity more rapidly, before reaching the Pauli limit, which strongly suggests the presence of Ising pairing. First-principles calculations and symmetry analysis confirm significant valley-selective spin splitting with out-of-plane spin polarization, further supporting the emergence of Ising superconductivity in 4$H$-NbSe$_2$. These results establish 4$H$-NbSe$_2$ as a robust bulk platform to investigate Ising superconductivity and valley-selective phenomena in transition-metal dichalcogenides.

cond-mat.supr-con↗

arXiv2025

Classifying Unreliable Narrators with Large Language Models

Often when we interact with a first-person account of events, we consider whether or not the narrator, the primary speaker of the text, is reliable. In this paper, we propose using computational methods to identify unreliable narrators, i.e. those who unintentionally misrepresent information. Borrowing literary theory from narratology to define different types of unreliable narrators based on a variety of textual phenomena, we present TUNa, a human-annotated dataset of narratives from multiple domains, including blog posts, subreddit posts, hotel reviews, and works of literature. We define classification tasks for intra-narrational, inter-narrational, and inter-textual unreliabilities and analyze the performance of popular open-weight and proprietary LLMs for each. We propose learning from literature to perform unreliable narrator classification on real-world text data. To this end, we experiment with few-shot, fine-tuning, and curriculum learning settings. Our results show that this task is very challenging, and there is potential for using LLMs to identify unreliable narrators. We release our expert-annotated dataset and code and invite future research in this area.

cs.CL↗

arXiv2025

MarginSel : Max-Margin Demonstration Selection for LLMs

Large Language Models (LLMs) excel at few-shot learning via in-context learning (ICL). However, the effectiveness of ICL is often sensitive to the selection and ordering of demonstration examples. To address this, we present MarginSel: Max-Margin Demonstration Selection for LLMs, a two-step method that selects hard demonstration examples for the ICL prompt, adapting to each test instance. Our approach achieves 2-7% absolute improvement in F1-score across classification tasks, compared to a random selection of examples. We also provide theoretical insights and empirical evidence showing that MarginSel induces max-margin behavior in LLMs by effectively increasing the margin for hard examples, analogous to support vectors, thereby shifting the decision boundary in a beneficial direction.

cs.LG↗

arXiv2025

INTERACT: Enabling Interactive, Question-Driven Learning in Large Language Models

Large language models (LLMs) excel at answering questions but remain passive learners-absorbing static data without the ability to question and refine knowledge. This paper explores how LLMs can transition to interactive, question-driven learning through student-teacher dialogues. We introduce INTERACT (INTERactive learning for Adaptive Concept Transfer), a framework in which a "student" LLM engages a "teacher" LLM through iterative inquiries to acquire knowledge across 1,347 contexts, including song lyrics, news articles, movie plots, academic papers, and images. Our experiments show that across a wide range of scenarios and LLM architectures, interactive learning consistently enhances performance, achieving up to a 25% improvement, with 'cold-start' student models matching static learning baselines in as few as five dialogue turns. Interactive setups can also mitigate the disadvantages of weaker teachers, showcasing the robustness of question-driven learning.

cs.CL↗

arXiv2025

Explicit Codes approaching Generalized Singleton Bound using Expanders

We construct a new family of explicit codes that are list decodable to capacity and achieve an optimal list size of $O(\frac{1}ε)$. In contrast to existing explicit constructions of codes achieving list decoding capacity, our arguments do not rely on algebraic structure but utilize simple combinatorial properties of expander graphs. Our construction is based on a celebrated distance amplification procedure due to Alon, Edmonds, and Luby [FOCS'95], which transforms any high-rate code into one with near-optimal rate-distance tradeoff. We generalize it to show that the same procedure can be used to transform any high-rate code into one that achieves list decoding capacity. Our proof can be interpreted as a "local-to-global" phenomenon for (a slight strengthening of) the generalized Singleton bound. Using this construction, for every $R, ε\in (0,1)$ and $k \in \mathbb{N}^+$, we obtain an \emph{explicit} family of codes $\mathcal{C} \subseteq Σ^n$, with rate $R$ such that, - They achieve the $ε$-relaxed generalized Singleton bound: for any $g \in Σ^n$ and any list $\mathcal{H}$ of at most $k$ codewords, we have, \[ \underset{h \in \mathcal{H}}{\mathbb{E}} [Δ(g,h)] ~\geq~ \frac{|\mathcal{H}|-1}{|\mathcal{H}|} \cdot (1 - R - ε). \] - The alphabet size is a constant depending only on $ε$ and $k$. - They can be list decoded up to radius $\frac{k-1}{k}(1-R-ε)$, in time $n^{O_{k,ε}(1)}$. As a corollary of our result, we also obtain the first explicit construction of LDPC codes achieving list decoding capacity, and in fact arbitrarily close to the generalized Singleton bound.

cs.IT↗

arXiv2025

Anomalous Magneto-transport and Anisotropic Multigap Superconductivity in Architecturally Misfit Layered System (PbS)$_{1.13}$TaS$_2$

Misfit-layered compounds, naturally occurring bulk heterostructures, present a compelling alternative to artificially engineered ones, offering a unique platform for exploring correlated phases and quantum phenomena. This study investigates the magnetotransport and superconducting properties of the misfit compound (PbS)$_{1.13}$TaS$_2$, comprising alternating PbS and 1$H$-TaS$_2$ layers. It exhibits distinctive transport properties, including a prominent planar Hall effect and a four-fold oscillatory Butterfly-shaped anisotropic magnetoresistance (AMR). Moreover, it shows multigap two-dimensional superconductivity with an exceptionally high in-plane upper critical field, exceeding the Pauli limit. The coexistence of unconventional superconductivity and anomalous transport - two distinct quantum phenomena, within the same material, suggests that misfit compounds provide an ideal platform for realizing quantum effects in the two-dimensional limit of bulk crystals. This opens the door to the development of simpler and more efficient quantum devices.

cond-mat.supr-con↗

arXiv2024

List Decodable Quantum LDPC Codes

We give a construction of Quantum Low-Density Parity Check (QLDPC) codes with near-optimal rate-distance tradeoff and efficient list decoding up to the Johnson bound in polynomial time. Previous constructions of list decodable good distance quantum codes either required access to a classical side channel or were based on algebraic constructions that preclude the LDPC property. Our construction relies on new algorithmic results for codes obtained via the quantum analog of the distance amplification scheme of Alon, Edmonds, and Luby [FOCS 1995]. These results are based on convex relaxations obtained using the Sum-of-Squares hierarchy, which reduce the problem of list decoding the distance amplified codes to unique decoding the starting base codes. Choosing these base codes to be the recent breakthrough constructions of good QLDPC codes with efficient unique decoders, we get efficiently list decodable QLDPC codes.

cs.IT↗

arXiv2024

DISCERN: Decoding Systematic Errors in Natural Language for Text Classifiers

Despite their high predictive accuracies, current machine learning systems often exhibit systematic biases stemming from annotation artifacts or insufficient support for certain classes in the dataset. Recent work proposes automatic methods for identifying and explaining systematic biases using keywords. We introduce DISCERN, a framework for interpreting systematic biases in text classifiers using language explanations. DISCERN iteratively generates precise natural language descriptions of systematic errors by employing an interactive loop between two large language models. Finally, we use the descriptions to improve classifiers by augmenting classifier training sets with synthetically generated instances or annotated examples via active learning. On three text-classification datasets, we demonstrate that language explanations from our framework induce consistent performance improvements that go beyond what is achievable with exemplars of systematic bias. Finally, in human evaluations, we show that users can interpret systematic biases more effectively (by over 25% relative) and efficiently when described through language explanations as opposed to cluster exemplars.

cs.CL↗

arXiv2024

SocialGaze: Improving the Integration of Human Social Norms in Large Language Models

While much research has explored enhancing the reasoning capabilities of large language models (LLMs) in the last few years, there is a gap in understanding the alignment of these models with social values and norms. We introduce the task of judging social acceptance. Social acceptance requires models to judge and rationalize the acceptability of people's actions in social situations. For example, is it socially acceptable for a neighbor to ask others in the community to keep their pets indoors at night? We find that LLMs' understanding of social acceptance is often misaligned with human consensus. To alleviate this, we introduce SocialGaze, a multi-step prompting framework, in which a language model verbalizes a social situation from multiple perspectives before forming a judgment. Our experiments demonstrate that the SocialGaze approach improves the alignment with human judgments by up to 11 F1 points with the GPT-3.5 model. We also identify biases and correlations in LLMs in assigning blame that is related to features such as the gender (males are significantly more likely to be judged unfairly) and age (LLMs are more aligned with humans for older narrators).

cs.CL↗

arXiv2024

Improved List Size for Folded Reed-Solomon Codes

Folded Reed-Solomon (FRS) codes are variants of Reed-Solomon codes, known for their optimal list decoding radius. We show explicit FRS codes with rate $R$ that can be list decoded up to radius $1-R-ε$ with lists of size $\mathcal{O}(1/ ε^2)$. This improves the best known list size among explicit list decoding capacity achieving codes. We also show a more general result that for any $k\geq 1$, there are explicit FRS codes with rate $R$ and distance $1-R$ that can be list decoded arbitrarily close to radius $\frac{k}{k+1}(1-R)$ with lists of size $(k-1)^2+1$. Our results are based on a new and simple combinatorial viewpoint of the intersections between Hamming balls and affine subspaces that recovers previously known parameters. We then use folded Wronskian determinants to carry out an inductive proof that yields sharper bounds.

cs.IT↗

arXiv2024

Fuse to Forget: Bias Reduction and Selective Memorization through Model Fusion

Model fusion research aims to aggregate the knowledge of multiple individual models to enhance performance by combining their weights. In this work, we study the inverse problem: investigating whether model fusion can be used to reduce unwanted knowledge. We investigate the effects of model fusion in three scenarios: the learning of shortcuts, social biases, and memorization of training data in fine-tuned language models. Through experiments covering classification and generation tasks, our analysis highlights that shared knowledge among models is enhanced during model fusion, while unshared knowledge is usually forgotten. Based on this observation, we demonstrate the potential of model fusion as a debiasing tool and showcase its efficacy in addressing privacy concerns associated with language models.

cs.CL↗

arXiv2024

Continuous Optimization for Decoding Errors

Error-correcting codes are one of the most fundamental objects in pseudorandomness, with applications in communication, complexity theory, and beyond. Codes are useful because of their ability to support decoding, which is the task of recovering a codeword from its noisy copy. List decoding is a relaxation where the decoder is allowed to output a list of codewords, and has seen tremendous progress over the last 25 years. In this thesis, we prove new algorithmic and combinatorial results about list decoding. We describe a list decoding framework that reduces the task of efficient decoding to proving distance in certain restricted proof systems. We then instantiate this framework for Tanner codes of Sipser and Spielman [IEEE Trans. Inf. Theory 1996] and Alon-Edmonds-Luby (AEL) distance amplification [FOCS 1995] of unique decodable base codes to get the first polynomial time list decoding algorithms for these codes. We also show extensions to the quantum version of AEL distance amplification to get polynomial-time list decodable quantum LDPC codes. We next give an alternate viewpoint of the above framework based on abstract regularity lemmas. We show how to efficiently implement the regularity lemma for the case of Ta-Shma's explicit codes near the Gilbert-Varshamov bound [STOC 2017]. This leads to a near-linear time algorithm for unique decoding of Ta-Shma's code. We also give new combinatorial results that improve known list sizes beyond the Johnson bound. Firstly, we adapt the AEL amplification to construct a new family of explicit codes that can be combinatorially list decoded to the optimal error correction radius. This is the first example of such a code that does not use significant algebraic ingredients. Secondly, we present list size improvements for Folded Reed-Solomon codes, improving the state of the art list size for explicit list decoding capacity achieving codes.

cs.IT↗

arXiv2023

Leveraging Multiple Teachers for Test-Time Adaptation of Language-Guided Classifiers

Recent approaches have explored language-guided classifiers capable of classifying examples from novel tasks when provided with task-specific natural language explanations, instructions or prompts (Sanh et al., 2022; R. Menon et al., 2022). While these classifiers can generalize in zero-shot settings, their task performance often varies substantially between different language explanations in unpredictable ways (Lu et al., 2022; Gonen et al., 2022). Also, current approaches fail to leverage unlabeled examples that may be available in many scenarios. Here, we introduce TALC, a framework that uses data programming to adapt a language-guided classifier for a new task during inference when provided with explanations from multiple teachers and unlabeled test examples. Our results show that TALC consistently outperforms a competitive baseline from prior work by an impressive 9.3% (relative improvement). Further, we demonstrate the robustness of TALC to variations in the quality and quantity of provided explanations, highlighting its potential in scenarios where learning from multiple teachers or a crowd is involved. Our code is available at: https://github.com/WeiKangda/TALC.git.

cs.CL↗

arXiv2023

A Comparison of Lexicon-Based and ML-Based Sentiment Analysis: Are There Outlier Words?

Lexicon-based approaches to sentiment analysis of text are based on each word or lexical entry having a pre-defined weight indicating its sentiment polarity. These are usually manually assigned but the accuracy of these when compared against machine leaning based approaches to computing sentiment, are not known. It may be that there are lexical entries whose sentiment values cause a lexicon-based approach to give results which are very different to a machine learning approach. In this paper we compute sentiment for more than 150,000 English language texts drawn from 4 domains using the Hedonometer, a lexicon-based technique and Azure, a contemporary machine-learning based approach which is part of the Azure Cognitive Services family of APIs which is easy to use. We model differences in sentiment scores between approaches for documents in each domain using a regression and analyse the independent variables (Hedonometer lexical entries) as indicators of each word's importance and contribution to the score differences. Our findings are that the importance of a word depends on the domain and there are no standout lexical entries which systematically cause differences in sentiment scores.

cs.CL↗

arXiv2023

Pragmatic Reasoning Unlocks Quantifier Semantics for Foundation Models

Generalized quantifiers (e.g., few, most) are used to indicate the proportions predicates are satisfied (for example, some apples are red). One way to interpret quantifier semantics is to explicitly bind these satisfactions with percentage scopes (e.g., 30%-40% of apples are red). This approach can be helpful for tasks like logic formalization and surface-form quantitative reasoning (Gordon and Schubert, 2010; Roy et al., 2015). However, it remains unclear if recent foundation models possess this ability, as they lack direct training signals. To explore this, we introduce QuRe, a crowd-sourced dataset of human-annotated generalized quantifiers in Wikipedia sentences featuring percentage-equipped predicates. We explore quantifier comprehension in language models using PRESQUE, a framework that combines natural language inference and the Rational Speech Acts framework. Experimental results on the HVD dataset and QuRe illustrate that PRESQUE, employing pragmatic reasoning, performs 20% better than a literal reasoning baseline when predicting quantifier percentage scopes, with no additional training required.

cs.AI↗

arXiv2023

List Decoding of Tanner and Expander Amplified Codes from Distance Certificates

We develop new list decoding algorithms for Tanner codes and distance-amplified codes based on bipartite spectral expanders. We show that proofs exhibiting lower bounds on the minimum distance of these codes can be used as certificates discoverable by relaxations in the Sum-of-Squares (SoS) semidefinite programming hierarchy. Combining these certificates with certain entropic proxies to ensure that the solutions to the relaxations cover the entire list, then leads to algorithms for list decoding several families of codes up to the Johnson bound. We prove the following: - We show that the LDPC Tanner codes of Sipser-Spielman [IEEE Trans. Inf. Theory 1996] and Zémor [IEEE Trans. Inf. Theory 2001] with alphabet size $q$, block-length $n$ and distance $δ$, based on an expander graph with degree $d$, can be list-decoded up to distance $\mathcal{J}_q(δ) - ε$ in time $n^{O_{d,q}(1/ε^4)}$, where $\mathcal{J}_q(δ)$ denotes the Johnson bound. - We show that the codes obtained via the expander-based distance amplification procedure of Alon, Edmonds and Luby [FOCS 1995] can be list-decoded close to the Johnson bound using the SoS hierarchy, by reducing the list decoding problem to unique decoding of the base code. In particular, starting from \emph{any} base code unique-decodable up to distance $δ$, one can obtain near-MDS codes with rate $R$ and distance $1-R - ε$, list-decodable up to the Johnson bound in time $n^{O_{ε, δ}(1)}$. - We show that the locally testable codes of Dinur et al. [STOC 2022] with alphabet size $q$, block-length $n$ and distance $δ$ based on a square Cayley complex with generator sets of size $d$, can be list-decoded up to distance $\mathcal{J}_q(δ) - ε$ in time $n^{O_{d,q}(1/ε^{4})}$, where $\mathcal{J}_q(δ)$ denotes the Johnson bound.

cs.DS↗