arXiv subjects

Sabyasachi Basu

Publications and source records attributed to Sabyasachi Basu.

11 recordsLinked to original sources

arXiv2026

EigenLI: Spectral Approximations to Late Interaction

Late-interaction models such as ColBERT achieve strong effectiveness by representing each document with many token-level vectors, but this expressivity leads to large indexing cost, storage footprints and expensive MaxSim scoring. We show that late-interaction representations exhibit an intrinsic low-rank structure: document token embeddings concentrate in a low-dimensional subspace that preserves most of the retrieval signal. Leveraging this observation, we introduce EigenLI, a spectral approximation framework that compresses late-interaction representations via document-specific low-dimensional subspaces. Unlike clustering or pooling methods, EigenLI identifies the dominant eigendirections of each document and uses them to construct reduced interaction representations. Empirically, $k$-EigenLI with $k \le 32$ outperforms k-means and Ward clustering based pooling methods on ColBERTv2 and AnswerAI-ColBERT-small; GTE-ModernColBERT exhibits a different tradeoff at $k=32$, where clustering methods perform better. The same spectral construction also yields EigenLI-SV, an ANN-compatible single-vector representation derived from the second-order summary of the reduced structure. Across multiple datasets and all three text models, EigenLI-SV consistently outperforms comparable single-vector surrogates such as MUVERA.

cs.IR↗

arXiv2026

Finding Many Overlapping Dense Subgraphs Using Triadic Cohorts

Graphs are a standard representation for data in the social sciences, cybersecurity, computer infrastructure, bioinformatics, and more. Typical real-world graphs are sparse, meaning the average degree is small (in the tens, while the number of vertices is more than millions). When graph data is collected from a source, a major task is to perform data exploration. Thus, any region of ``density" is of interest, since it indicates special structure. An important goal is to cover a significant portion of the graph using dense subgraphs. Existing algorithms that find many dense subgraphs do not have overlapping output, and hence provide limited coverage. Other methods that produce overlapping clusters do not generate dense subgraphs. The main goal of this paper is to develop provable and practical methods that can provide overlapping dense subgraphs that can cover large portions of real-world networks. Our contribution is an algorithm CohortRecovery that achieves this goal. We first develop a mathematical framework of triadic cohorts that captures the notion of ``detectable dense subgraphs'' that potentially overlap. We prove that CohortRecovery can output a set of dense subgraphs, such that each triadic cohort is almost completely contained in some dense subgraph. We give a practical implementation of CohortRecovery and demonstrate it on a variety of datasets. It typically runs in under ten minutes on a commodity machine even on graphs with tens of millions of edges. For numerous datasets, CohortRecovery is able to cover more than 25\% of the vertices in non-trivial subgraphs of density more than 0.8, and is significantly better than a wide variety of scalable graph clustering/community detection algorithms. Moreover, we demonstrate that output of CohortRecovery captures ground truth clusters obtained by manual curation.

cs.SI↗

arXiv2026

Retrieval Needs Multivectors: An Exponential Separation

Recent works have highlighted the expressive limitations of embedding based retrieval models through both theoretical analyses and challenging benchmarks such as LIMIT. While multi-vector embeddings consistently outperform single-vector embeddings, the precise representational gap between them remains poorly understood. In this work, following Jayaram's work, we provide the first explicit family of query and document sets, together with their relevance matrices, for which single-vector embeddings that rank all relevant documents above irrelevant ones require exponential size, whereas polynomial-size multi-vector embeddings suffice. Our result establishes an exponential separation between the expressive power of single-vector and multi-vector embeddings for the task of ranking of documents as opposed to approximating numerical scores as in the work of Jayaram. Motivated by our theoretical construction, we introduce ANDOR, a new retrieval benchmark that naturally instantiates these hard examples. We show that state-of-the-art single-vector embedding models perform poorly on ANDOR in the zero-shot setting and exhibit only marginal improvements after fine-tuning, highlighting the inherent difficulty of the benchmark compared to prior work. In contrast, multi-vector models consistently outperform their single-vector counterparts and improve substantially with fine-tuning, closely aligning with our theoretical predictions.

cs.IR↗

arXiv2026

Quantization Beyond Uniform Bit Allocation

Quantization is a fundamental technique to handle the growing sizes of embeddings generated by modern models. Existing quantization schemes are largely embedding agnostic and allocate bits uniformly across dimensions. However, recent models produce embeddings with significant geometric structure. In this work, we investigate whether a variable bit allocation scheme can improve quantization quality under a fixed memory budget. We propose a simple variable bit allocation framework that partitions an embedding into contiguous buckets and allocates storage non-uniformly across them. Using a greedy allocation strategy, we instantiate this framework for both Product Quantization (PQ) and Scalar Quantization (SQ). We perform a series of experiments on embeddings known to have the Matryoshka property (MRL), and consistently observe that non-uniform allocations outperform uniform baselines at identical storage budgets. The largest improvements occur in the low-bit regime, where uniform allocation is particularly inefficient for MRL embeddings. At the same compression rates, variable allocation improves recall by up to 8\% for PQ and up to 18\% for SQ. Our results suggest a new direction for structure-aware compression and indexing techniques for large-scale retrieval systems.

cs.IR↗

arXiv2026

Query-Limited Community Recovery in Stochastic Block Models

We study exact community recovery in the two-community stochastic block model on $n$ vertices under limited and noisy access to network data. The learner may query a noisy neighborhood oracle that reveals each true neighbor of a queried vertex independently with fixed probability and never returns non-neighbors, subject to a finite query budget. We consider both oracle-only access and a combined model where the learner also observes a single subsampled copy of the underlying graph. For oracle-only access, balanced uniform querying gives a sharp non-adaptive benchmark: when each vertex is queried the same integer number of times, the observations reduce to an SBM with attenuated edge probabilities and the Abbe-Bandeira-Hall exact-recovery threshold applies. We show that this benchmark is not adaptively optimal: a two-stage adaptive strategy succeeds with $n+o(n)$ queries in a regime where balanced uniform querying requires $m n$ queries for some $m>1$. With an additional subsampled graph, we prove a sublinear-query adaptivity gap: balanced data-independent uniform querying with a sublinear budget does not improve over the subsampled graph alone, whereas adaptive querying can target a small set of uncertain vertices and achieve exact recovery. Thus adaptive data acquisition can strictly improve the information-theoretic limits of exact recovery.

cs.IT↗

arXiv2025

Aggregating maximal cliques in real-world graphs

Maximal clique enumeration is a fundamental graph mining task, but its utility is often limited by computational intractability and highly redundant output. To address these challenges, we introduce \emph{$\rho$-dense aggregators}, a novel approach that succinctly captures maximal clique structure. Instead of listing all cliques, we identify a small collection of clusters with edge density at least $\rho$ that collectively contain every maximal clique. In contrast to maximal clique enumeration, we prove that for all $\rho < 1$, every graph admits a $\rho$-dense aggregator of \emph{sub-exponential} size, $n^{O(\log_{1/\rho}n)}$, and provide an algorithm achieving this bound. For graphs with bounded degeneracy, a typical characteristic of real-world networks, our algorithm runs in near-linear time and produces near-linear size aggregators. We also establish a matching lower bound on aggregator size, proving our results are essentially tight. In an empirical evaluation on real-world networks, we demonstrate significant practical benefits for the use of aggregators: our algorithm is consistently faster than the state-of-the-art clique enumeration algorithm, with median speedups over $6\times$ for $\rho=0.1$ (and over $300\times$ in an extreme case), while delivering a much more concise structural summary.

cs.DS↗

arXiv2024

Covering a Graph with Dense Subgraph Families, via Triangle-Rich Sets

Graphs are a fundamental data structure used to represent relationships in domains as diverse as the social sciences, bioinformatics, cybersecurity, the Internet, and more. One of the central observations in network science is that real-world graphs are globally sparse, yet contains numerous "pockets" of high edge density. A fundamental task in graph mining is to discover these dense subgraphs. Most common formulations of the problem involve finding a single (or a few) "optimally" dense subsets. But in most real applications, one does not care for the optimality. Instead, we want to find a large collection of dense subsets that covers a significant fraction of the input graph. We give a mathematical formulation of this problem, using a new definition of regularly triangle-rich (RTR) families. These families capture the notion of dense subgraphs that contain many triangles and have degrees comparable to the subgraph size. We design a provable algorithm, RTRExtractor, that can discover RTR families that approximately cover any RTR set. The algorithm is efficient and is inspired by recent results that use triangle counts for community testing and clustering. We show that RTRExtractor has excellent behavior on a large variety of real-world datasets. It is able to process graphs with hundreds of millions of edges within minutes. Across many datasets, RTRExtractor achieves high coverage using high edge density datasets. For example, the output covers a quarter of the vertices with subgraphs of edge density more than (say) $0.5$, for datasets with 10M+ edges. We show an example of how the output of RTRExtractor correlates with meaningful sets of similar vertices in a citation network, demonstrating the utility of RTRExtractor for unsupervised graph discovery tasks.

cs.SI↗

arXiv2024

A Sublinear Algorithm for Approximate Shortest Paths in Large Networks

Computing distances and finding shortest paths in massive real-world networks is a fundamental algorithmic task in network analysis. There are two main approaches to solving this task. On one hand are traversal-based algorithms like bidirectional breadth-first search (BiBFS) with no preprocessing step and slow individual distance inquiries. On the other hand are indexing-based approaches, which maintain a large index. This allows for answering individual inquiries very fast; however, index creation is prohibitively expensive. We seek to bridge these two extremes: quickly answer distance inquiries without the need for costly preprocessing. In this work, we propose a new algorithm and data structure, WormHole, for approximate shortest path computations. WormHole leverages structural properties of social networks to build a sublinearly sized index, drawing upon the explicit core-periphery decomposition of Ben-Eliezer et al. Empirically, the preprocessing time of WormHole improves upon index-based solutions by orders of magnitude, and individual inquiries are consistently much faster than in BiBFS. The acceleration comes at the cost of a minor accuracy trade-off. Nonetheless, our empirical evidence demonstrates that WormHole accurately answers essentially all inquiries within a maximum additive error of 2. We complement these empirical results with provable theoretical guarantees, showing that WormHole requires $n^{o(1)}$ node queries per distance inquiry in random power-law networks. In contrast, any approach without a preprocessing step requires $n^{Ω(1)}$ queries for the same task. WormHole does not require reading the whole graph. Unlike the vast majority of index-based algorithms, it returns paths, not just distances. For faster inquiry times, it can be combined effectively with other index-based solutions, by running them only on the sublinear core.

cs.DS↗

arXiv2024

Spectral Triadic Decompositions of Real-World Networks

A fundamental problem in mathematics and network analysis is to find conditions under which a graph can be partitioned into smaller pieces. The most important tool for this partitioning is the Fiedler vector or discrete Cheeger inequality. These results relate the graph spectrum (eigenvalues of the normalized adjacency matrix) to the ability to break a graph into two pieces, with few edge deletions. An entire subfield of mathematics, called spectral graph theory, has emerged from these results. Yet these results do not say anything about the rich community structure exhibited by real-world networks, which typically have a significant fraction of edges contained in numerous densely clustered blocks. Inspired by the properties of real-world networks, we discover a new spectral condition that relates eigenvalue powers to a network decomposition into densely clustered blocks. We call this the \emph{spectral triadic decomposition}. Our relationship exactly predicts the existence of community structure, as commonly seen in real networked data. Our proof provides an efficient algorithm to produce the spectral triadic decomposition. We observe on numerous social, coauthorship, and citation network datasets that these decompositions have significant correlation with semantically meaningful communities.

cs.SI↗

arXiv2023

Optimal storage codes on graphs with fixed locality

Storage codes on graphs are an instance of \emph{codes with locality}, which are used in distributed storage schemes to provide local repairability. Specifically, the nodes of the graph correspond to storage servers, and the neighbourhood of each server constitute the set of servers it can query to repair its stored data in the event of a failure. A storage code on a graph with $n$-vertices is a set of $n$-length codewords over $\field_q$ where the $i$th codeword symbol is stored in server $i$, and it can be recovered by querying the neighbours of server $i$ according to the underlying graph. In this work, we look at binary storage codes whose repair function is the parity check, and characterise the tradeoff between the locality of the code and its rate. Specifically, we show that the maximum rate of a code on $n$ vertices with locality $r$ is bounded between $1-1/n\lceil n/(r+1)\rceil$ and $1-1/n\lceil n/(r+1)\rceil$. The lower bound on the rate is derived by constructing an explicit family of graphs with locality $r$, while the upper bound is obtained via a lower bound on the binary-field rank of a class of symmetric binary matrices. Our upper bound on maximal rate of a storage code matches the upper bound on the larger class of codes with locality derived by Tamo and Barg. As a corollary to our result, we obtain the following asymptotic separation result: given a sequence $r(n), n\geq 1$, there exists a sequence of graphs on $n$-vertices with storage codes of rate $1-o(1)$ if and only if $r(n)=ω(1)$.

cs.IT↗

arXiv2021

The complexity of testing all properties of planar graphs, and the role of isomorphism

Consider property testing on bounded degree graphs and let $\varepsilon>0$ denote the proximity parameter. A remarkable theorem of Newman-Sohler (SICOMP 2013) asserts that all properties of planar graphs (more generally hyperfinite) are testable with query complexity only depending on $\varepsilon$. Recent advances in testing minor-freeness have proven that all additive and monotone properties of planar graphs can be tested in $poly(\varepsilon^{-1})$ queries. Some properties falling outside this class, such as Hamiltonicity, also have a similar complexity for planar graphs. Motivated by these results, we ask: can all properties of planar graphs can be tested in $poly(\varepsilon^{-1})$ queries? Is there a uniform query complexity upper bound for all planar properties, and what is the "hardest" such property to test? We discover a surprisingly clean and optimal answer. Any property of bounded degree planar graphs can be tested in $\exp(O(\varepsilon^{-2}))$ queries. Moreover, there is a matching lower bound, up to constant factors in the exponent. The natural property of testing isomorphism to a fixed graph needs $\exp(Ω(\varepsilon^{-2}))$ queries, thereby showing that (up to polynomial dependencies) isomorphism to an explicit fixed graph is the hardest property of planar graphs. The upper bound is a straightforward adapation of the Newman-Sohler analysis that tracks dependencies on $\varepsilon$ carefully. The main technical contribution is the lower bound construction, which is achieved by a special family of planar graphs that are all mutually far from each other. We can also apply our techniques to get analogous results for bounded treewidth graphs. We prove that all properties of bounded treewidth graphs can be tested in $\exp(O(\varepsilon^{-1}\log \varepsilon^{-1}))$ queries. Moreover, testing isomorphism to a fixed forest requires $\exp(Ω(\varepsilon^{-1}))$ queries.

cs.DS↗