arXiv subjects

Siddharth Pal

Publications and source records attributed to Siddharth Pal.

14 recordsLinked to original sources

arXiv2026

Quantum-enhanced physical-layer threat detection in metropolitan-scale fiber networks

Network security is widely recognized as a key application of quantum technology. However, its large-scale deployment is hindered by the need for tight coordination between fundamentally different quantum and classical processing steps in conventional protocols. This requirement introduces strong cross-layer interdependencies that conflict with the modular, layered architectures enabling scalability in modern communication networks. Here, we present an alternative strategy that confines all quantum interventions to the physical layer and remains transparently compatible with existing network abstractions. This is achieved by directly embedding quantum features and classical information within the same optical field using bright squeezed light. Physical-layer signals are analyzed using a cumulative sum (CUSUM) method to enable quantum-enhanced threat detection. We validate the practicality of this approach through field deployment over a metropolitan-scale fiber network and further demonstrate network-level security functionalities enabled by physical-layer quantum-enhanced thread detection. These results establish a practical, scalable framework for seamlessly integrating quantum-enhanced security into large-scale communication infrastructure.

quant-ph↗

arXiv2026

Context by Distinct Information: An Auditable Dirichlet-Process Working Memory for Long, Redundant Context Streams

Context engineering decides what information a model carries forward, and current designs meter it in tokens: compressing the past into a bounded recurrent state, keeping a key-value entry for every token, or imposing a fixed budget through a window or eviction rule. All three make the token the unit of memory even when the stream is redundant and the task depends on the distinct information it carries. Building on a companion mechanism paper that opens a cache slot only when an incoming key is novel, so memory scales with the number of distinct items rather than tokens, we develop that allocate-on-novelty cache as a working-memory component and organize context by how a task depends on the past: recall-carried information belongs in a content-addressed novelty cache, summary-carried information in a recurrent state, and locality-carried information in a recency window. The claim is empirical and bounded. On a matched character-level control, novelty-gated attention reaches full-attention performance while attending to about half the tokens, and coupling the cache with a state-space summary matches full-attention coupling at that reduced cost; the advantage grows as context lengthens, while a sliding window is preferable on short, locality-dominated spans. On next-code prediction over synthetic Medicare claims the coupled component leads full attention and every fixed-budget eviction policy at a thousand-event horizon, whereas cost forecasting over the same stream is summary-carried and the cache is neutral. The retained memory is an inspectable table of templates, codes, drugs, or places rather than an opaque state. The experiments are small-scale and use only public data; they establish the primitive that context can scale with distinct information rather than tokens, in a working memory that is content-addressable and auditable.

cs.LG↗

arXiv2026

Remembering Distinct Items, Not Tokens: A Learnable Dirichlet-Process Cache Between State-Space Models and Attention

Fixed-state sequence models compress an unbounded past into a bounded state, which caps their associative recall at roughly the state dimension; attention escapes the cap by keeping a key-value entry for every token, at quadratic compute and a cache that grows with the sequence. We study the middle ground: a sparse cache that allocates a slot only when an input is novel, so its size tracks the number of distinct items rather than the number of tokens. The allocation rule is the DP-means clustering rule, the small-variance limit of a Dirichlet-process mixture, used not as latent-variable inference but as the key-value memory operator for a deep recurrent backbone. We develop it in two forms, a static cache with a fixed concentration and a surprise-adaptive variant whose concentration follows the recent novelty rate. On a controlled associative-recall benchmark with redundancy we show that the cache matches full-attention recall while storing only the distinct items, that it dominates a fixed-budget eviction cache on the recall-versus-size frontier, and that on a state-space backbone it answers both a recall query and a long-range aggregate at the lowest memory of any model tested. The allocation is learnable end to end: a two-parameter novelty-threshold gate trained on the task loss alone recovers the rule exactly, whereas an over-parameterized gate fails, so the operative ingredient is the inductive bias rather than capacity. The evidence is a family of controlled mechanism studies at modest scale, with the distinct-items property confirmed on four real streams (recommendation, systems logs, clinical events, and insurance claims); a real-backbone, real-corpus language validation is pursued in a companion study.

cs.LG↗

arXiv2026

Scaling Audio Models Efficiently: A Joint Study of Compute Constraints and Optimization Behavior

Large automatic speech recognition (ASR) models such as Whisper must be deployed across hardware with widely varying memory and inference-speed constraints. We present a compression framework that jointly parametrizes Whisper deployment along \emph{six} dimensions: model size $x_N$, temporal resolution $x_T$, encoder token stride $x_V$, low-rank adaptation capacity $x_R$, weight precision $x_Q$ and sparsity pattern $x_P$. All axes are jointly optimized against three deployment objectives (word error rate, inference FLOPs, and memory footprint) using a non-dominated sorting genetic evolutionary search (NSGA). Across 50 of the 1,680 candidate configurations evaluated, we measure the marginal effect of each axis on the three objectives and identify compression combinations that dominate naive single-axis scaling, and report a consistent negative result: 1:4 structured sparsity fails to recover acceptable accuracy under any tested recovery budget. We report real, measured memory and accuracy figures for genuinely quantized deployment artifacts, and provide a lookup table mapping deployment scenarios (cloud, server, edge, ultra-constrained) to specific axis configurations with their measured accuracy/memory/compute trade-offs

cs.SD↗

arXiv2026

A Stationarity-and-Coupling Criterion for Training-Free Time-Lagged Spectral Embeddings of Multivariate Time Series

We study training-free fixed-length descriptors for multivariate time series and ask not merely whether such a descriptor performs well, but when it can be expected to work at all. Our object of study is $D(\tau)$, built from a time-lagged correlation matrix truncated at the Marchenko-Pastur edge so that only signal-bearing eigenvalues survive and classified by cosine similarity to class centroids with zero learned parameters. The central contribution is not the descriptor but a falsifiable applicability criterion for it. Working from a stationary Gaussian VAR(1) model, we argue that $D(\tau)$ separates two classes when the signals are approximately stationary and the class information lives in their cross-channel temporal coupling rather than in marginal per-channel power. We derive, semi-formally, three consequences: a distinguishability condition, why the static ($\tau=0$) covariance collapses to chance, and why a stationary but power-discriminated paradigm defeats the descriptor. The criterion is operational: a two-part pre-flight test -- an augmented Dickey-Fuller stationarity check and a power-baseline saturation check -- predicts applicability before any training. We validate both halves on a mixed assortment. On four paradigms that satisfy the criterion (Sleep-EDF, BCI-IV-2a, MIT-BIH, ESC-50) the descriptor is competitive with strong baselines at a fraction of their cost, reaching $88.5\pm4.5\%$ under 20-subject leave-one-subject-out on Sleep-EDF on a single CPU thread. On three that violate it -- non-stationary ERPs, and financial-volatility and wearable-stress regimes that are power-discriminated -- it fails exactly as the pre-flight predicts, and these negatives are the more informative half. We are explicit that $D(\tau)$ is not the most accurate representation; its value is a compact, training-free embedding whose domain of validity is known in advance.

cs.LG↗

arXiv2024

Efficient All-to-All Collective Communication Schedules for Direct-Connect Topologies

The all-to-all collective communications primitive is widely used in machine learning (ML) and high performance computing (HPC) workloads, and optimizing its performance is of interest to both ML and HPC communities. All-to-all is a particularly challenging workload that can severely strain the underlying interconnect bandwidth at scale. This paper takes a holistic approach to optimize the performance of all-to-all collective communications on supercomputer-scale direct-connect interconnects. We address several algorithmic and practical challenges in developing efficient and bandwidth-optimal all-to-all schedules for any topology and lowering the schedules to various runtimes and interconnect technologies. We also propose a novel topology that delivers near-optimal all-to-all performance.

cs.DC↗

arXiv2022

Efficient Direct-Connect Topologies for Collective Communications

We consider the problem of distilling efficient network topologies for collective communications. We provide an algorithmic framework for constructing direct-connect topologies optimized for the latency vs. bandwidth trade-off associated with the workload. Our approach synthesizes many different topologies and schedules for a given cluster size and degree and then identifies the appropriate topology and schedule for a given workload. Our algorithms start from small, optimal base topologies and associated communication schedules and use techniques that can be iteratively applied to derive much larger topologies and schedules. Additionally, we incorporate well-studied large-scale graph topologies into our algorithmic framework by producing efficient collective schedules for them using a novel polynomial-time algorithm. Our evaluation uses multiple testbeds and large-scale simulations to demonstrate significant performance benefits from our derived topologies and schedules.

cs.NI↗

arXiv2020

Dynamics of node influence in network growth models

Many classes of network growth models have been proposed in the literature for capturing real-world complex networks. Existing research primarily focuses on global characteristics of these models, e.g., degree distribution. We aim to shift the focus towards studying the network growth dynamics from the perspective of individual nodes. In this paper, we study how a metric for node influence in network growth models behaves over time as the network evolves. This metric, which we call node visibility, captures the probability of the node to form new connections. First, we conduct an investigation on three popular network growth models -- preferential attachment, additive, and multiplicative fitness models; and primarily look into the "influential nodes" or "leaders" to understand how their visibility evolves over time. Subsequently, we consider a generic fitness model and observe that the multiplicative model strikes a balance between allowing influential nodes to maintain their visibility, while at the same time making it possible for new nodes to gain visibility in the network. Finally, we observe that a spatial growth model with multiplicative fitness can curtail the global reach of influential nodes, thereby allowing the emergence of a multiplicity of "local leaders" in the network.

cs.SI↗

arXiv2020

Modeling Citation Trajectories of Scientific Papers

Several network growth models have been proposed in the literature that attempt to incorporate properties of citation networks. Generally, these models aim at retaining the degree distribution observed in real-world networks. In this work, we explore whether existing network growth models can realize the diversity in citation growth exhibited by individual papers - a new node-centric property observed recently in citation networks across multiple domains of research. We theoretically and empirically show that the network growth models which are solely based on degree and/or intrinsic fitness cannot realize certain temporal growth behaviors that are observed in real-world citation networks. To this end, we propose two new growth models that localize the influence of papers through an appropriate attachment mechanism. Experimental results on the real-world citation networks of Computer Science and Physics domains show that our proposed models can better explain the temporal behavior of citation networks than existing models.

cs.SI↗

arXiv2019

Asymptotic degree distributions in random threshold graphs

We discuss several limiting degree distributions for a class of random threshold graphs in the many node regime. This analysis is carried out under a weak assumption on the distribution of the underlying fitness variable. This assumption, which is satisfied by the exponential distribution, determines a natural scaling under which the following limiting results are shown: The nodal degree distribution, i.e., the distribution of any node, converges in distribution to a limiting pmf. However, for each $d=0,1, \ldots $, the fraction of nodes with given degree $d$ converges only in distribution to a non-degenerate random variable $Π(d)$ (whose distribution depends on $d$),and not in probability to the aforementioned limiting nodal pmf as is customarily expected. The distribution of $Π(d)$ is identified only through its characteristic function. Implications of this result include: (i) The empirical node distribution may not be used as a proxy for or as an estimate to the limiting nodal pmf; (ii) Even in homogeneous graphs, the network-wide degree distribution and the nodal degree distribution may capture vastly different information; and (iii) Random threshold graphs with exponential distributed fitness do not provide an alternative scale-free model to the Barabási-Albert model as was argued by some authors; the two models cannot be meaningfully compared in terms of their degree distributions!

math.PR↗

arXiv2018

Asymptotic degree distributions in large (homogeneous) random networks: A little theory and a counterexample

In random graph models, the degree distribution of an individual node should be distinguished from the (empirical) degree distribution of the graph that records the fractions of nodes with given degree. We introduce a general framework to explore when these two degree distributions coincide asymptotically in large homogeneous random networks. The discussion is carried under three basic statistical assumptions on the degree sequences: (i) a weak form of distributional homogeneity; (ii) the existence of an asymptotic (nodal) degree distribution; and (iii) a weak form of asymptotic uncorrelatedness. We show that this asymptotic equality may fail in homogeneous random networks for which (i) and (ii) hold but (iii) does not. The counterexample is found in the class of random threshold graphs. An implication of this finding is that random threshold graphs cannot be used as a substitute to the Barabási-Albert model for scale-free network modeling, as has been proposed by some authors. The results can also be formulated for non-homogeneous models by making use of a random sampling procedure over the nodes.

cs.SI↗

arXiv2018

Quantum Walk Inspired Neural Networks for Graph-Structured Data

In recent years, new neural network architectures designed to operate on graph-structured data have pushed the state-of-the-art in the field. A large set of these architectures utilize a form of classical random walks to diffuse information. We propose quantum walk neural networks (QWNN), a novel graph neural network architecture based on quantum random walks, the quantum parallel to classical random walks. A QWNN learns a quantum walk on a graph to construct a diffusion operator which can then be applied to graph-structured data. We demonstrate the use of QWNNs on a variety of prediction tasks on graphs involving temperature, biological, and molecular datasets.

quant-ph↗

arXiv2017

Sparse Diffusion-Convolutional Neural Networks

The predictive power and overall computational efficiency of Diffusion-convolutional neural networks make them an attractive choice for node classification tasks. However, a naive dense-tensor-based implementation of DCNNs leads to $\mathcal{O}(N^2)$ memory complexity which is prohibitive for large graphs. In this paper, we introduce a simple method for thresholding input graphs that provably reduces memory requirements of DCNNs to O(N) (i.e. linear in the number of nodes in the input) without significantly affecting predictive performance.

cs.LG↗

arXiv2017

An efficient alternative to Ollivier-Ricci curvature based on the Jaccard metric

We study Ollivier-Ricci curvature, a discrete version of Ricci curvature, which has gained popularity over the past several years and has found applications in diverse fields. However, the Ollivier-Ricci curvature requires an optimal mass transport problem to be solved, which can be computationally expensive for large networks. In view of this, we propose two alternative measures of curvature to Ollivier-Ricci which are motivated by the Jaccard coefficient and are demonstrably less computationally intensive, a cheaper Jaccard (JC) and a more expensive generalized Jaccard (gJC) curvature metric. We show theoretically that the gJC closely matches the Ollivier-Ricci curvature for Erdos-Renyi graphs in the asymptotic regime of large networks. Furthermore, we study the goodness of approximation between the proposed curvature metrics and Ollivier-Ricci curvature for several network models and real networks. Our results suggest that in comparison to an alternative curvature metric for graphs, the Forman-Ricci curvature, the gJC exhibits a reasonably good fit to the Ollivier-Ricci curvature for a wide range of networks, while the JC is shown to be a good proxy only for certain scenarios.

cs.SI↗