arXiv subjects

Keshav Goyal

Publications and source records attributed to Keshav Goyal.

8 recordsLinked to original sources

arXiv2026

AnySimLite: A Lightweight Few-Shot Similarity Encoder for On-Device Speech-Adjacent Classification

To minimize privacy concerns and inference latency on edge devices like smartphones, lightweight on-device models remain important for end-user applications. Many of these applications involve natural language classification, but deploying multiple specialized models creates a memory footprint challenge. We investigate: Can a single lightweight architecture solve multiple Speech-Adjacent (SA) classification tasks through reduction to a nuanced text similarity formulation? We propose AnySimLite, a lightweight similarity encoder that combines word-level and character-level channels. Together with a dataset transformation strategy, we evaluate AnySimLite across multiple SA classification tasks and show that it consistently achieves state-of-the-art (SOTA) or SOTA-competitive performance in few-shot settings while maintaining a low memory footprint. Even in the worst case, the performance drop remains below 7% while using $<\frac{1}{250}^{\mathrm{th}}$ of the model size of the SOTA qLLaMA_LoRA-7B baseline.

cs.CL↗

arXiv2026

The Synthesis-Sequencing Channel for DNA-based Data Storage

We introduce and study the synthesis-sequencing channel, a two-stage model for DNA-based data storage that jointly captures synthesis and sequencing effects. The synthesis-sequencing channel provides a more nuanced and realistic model of the DNA storage process compared to prior work, as it distinguishes between physical coverage after synthesis and sequencing coverage after readout, relaxes the assumption of independent errors across reads, and naturally induces coverage bias through the composition of synthesis and sequencing stages. We establish the information-theoretic capacity of this channel by deriving matching converse and achievability bounds for the case where synthesis and sequencing errors are modeled by binary symmetric channels with possibly different error probabilities, under mild assumptions on the channel parameters. Our results reveal multiple trade-offs between physical coverage, synthesis errors, sequencing coverage, and sequencing errors that influence the maximum achievable rate for reliable data storage.

cs.IT↗

arXiv2025

Learning Distinguishable Representations in Deep Q-Networks for Linear Transfer

Deep Reinforcement Learning (RL) has demonstrated success in solving complex sequential decision-making problems by integrating neural networks with the RL framework. However, training deep RL models poses several challenges, such as the need for extensive hyperparameter tuning and high computational costs. Transfer learning has emerged as a promising strategy to address these challenges by enabling the reuse of knowledge from previously learned tasks for new, related tasks. This avoids the need for retraining models entirely from scratch. A commonly used approach for transfer learning in RL is to leverage the internal representations learned by the neural network during training. Specifically, the activations from the last hidden layer can be viewed as refined state representations that encapsulate the essential features of the input. In this work, we investigate whether these representations can be used as input for training simpler models, such as linear function approximators, on new tasks. We observe that the representations learned by standard deep RL models can be highly correlated, which limits their effectiveness when used with linear function approximation. To mitigate this problem, we propose a novel deep Q-learning approach that introduces a regularization term to reduce positive correlations between feature representation of states. By leveraging these reduced correlated features, we enable more effective use of linear function approximation in transfer learning. Through experiments and ablation studies on standard RL benchmarks and MinAtar games, we demonstrate the efficacy of our approach in improving transfer learning performance and thereby reducing computational overhead.

cs.LG↗

arXiv2025

Bounds on Codes Correcting Adjacent Transpositions

We study the problem of correcting pairwise disjoint adjacent transpositions (or swaps) in $q$-ary strings. Equivalently, the model we assume is the radius-one instance of the so-called $\ell_\infty$-limited permutation channel. We first study the relevant combinatorial properties of the appropriately defined transposition distance, including center-specific and average ball sizes. We then derive two lower bounds and one upper bound on the asymptotic rates of optimal codes correcting $t=\tau n$ transpositions. The first achievability result is a generalized Gilbert--Varshamov bound, while the second follows from a construction of codes correcting all possible patterns of adjacent transpositions and therefore represents a lower bound on the zero-error capacity of this model. This construction improves the classical general-alphabet construction for $3\leqslant q\leqslant 8$ as well as the recent bounds for $q=4,5$. The upper bound is obtained by a packing argument adjusted to the run-structure of a given code. To the best of our knowledge, these are the first nonconstant, $\tau$-dependent lower and upper bounds developed for the pairwise disjoint $q$-ary model throughout the linear regime. We also derive asymptotic bounds on the cardinality of optimal codes correcting $t=\textrm{const}$ pairwise disjoint adjacent transpositions.

cs.IT↗

arXiv2024

Auctioning Escape Permits for Multiple Correlated Pollutants Using CMRA

In the context of increasingly complex environmental challenges, effective pollution control mechanisms are crucial. By extending the state of the art auction mechanisms, we aim to develop an efficient approach for allocating pollution abatement resources in a multi-pollutant setting with pollutants affecting each other's reduction costs. We modify the Combinatorial Multi-Round Ascending Auction for the auction of escape permits of pollutants with co-dependent reduction processes, specifically, greenhouse gas emissions and nutrient runoff in Finnish agriculture. We show the significant advantages of this mechanism in pollution control through experiments on the bid prices and amount of escape permits sold in multiple auction simulations.

cs.GT↗

arXiv2024

Evaluating the Gilbert-Varshamov Bound for Constrained Systems

We revisit the well-known Gilbert-Varshamov (GV) bound for constrained systems. In 1991, Kolesnik and Krachkovsky showed that GV bound can be determined via the solution of some optimization problem. Later, Marcus and Roth (1992) modified the optimization problem and improved the GV bound in many instances. In this work, we provide explicit numerical procedures to solve these two optimization problems and hence, compute the bounds. We then show the procedures can be further simplified when we plot the respective curves. In the case where the graph presentation comprise a single state, we provide explicit formulas for both bounds.

cs.IT↗

arXiv2024

Gilbert-Varshamov Bound for Codes in $L_1$ Metric using Multivariate Analytic Combinatorics

Analytic combinatorics in several variables refers to a suite of tools that provide sharp asymptotic estimates for certain combinatorial quantities. In this paper, we apply these tools to determine the Gilbert--Varshamov lower bound on the rate of optimal codes in $L_1$ metric. Several different code spaces are analyzed, including the simplex and the hypercube in $\mathbb{Z^n}$, all of which are inspired by concrete data storage and transmission models such as the sticky insertion channel, the permutation channel, the adjacent transposition (bit-shift) channel, the multilevel flash memory channel, etc.

cs.IT↗

arXiv2021

Sequence Reconstruction Problem for Deletion Channels: A Complete Asymptotic Solution

Transmit a codeword $x$, that belongs to an $(\ell-1)$-deletion-correcting code of length $n$, over a $t$-deletion channel for some $1\le \ell\le t<n$. Levenshtein, in 2001, proposed the problem of determining $N(n,\ell,t)+1$, the minimum number of distinct channel outputs required to uniquely reconstruct $x$. Prior to this work, $N(n,\ell,t)$ is known only when $\ell\in\{1,2\}$. Here, we provide an asymptotically exact solution for all values of $\ell$ and $t$. Specifically, we show that $N(n,\ell,t)=\binom{2\ell}{\ell}/(t-\ell)! n^{t-\ell} - O(n^{t-\ell-1})$ and in the special instance where $\ell=t$, we show that $N(n,\ell,\ell)=\binom{2\ell}{\ell}$. We also provide a conjecture on the exact value of $N(n,\ell,t)$ for all values of $n$, $\ell$, and $t$.

cs.IT↗