arXiv subjects

Lav R. Varshney

Publications and source records attributed to Lav R. Varshney.

At least 55 records · Page 3Linked to original sources

arXiv2025

Dynamic Batching of Online Arrivals to Leverage Economies of Scale

Many settings, such as matching riders to drivers in ride-hailing platforms or in-stream video advertising, require handling arrivals over time. In such applications, it is often beneficial to group the arriving orders or requests into batches and process the larger batches rather than individual arrivals. However, waiting too long to create larger batches incurs a waiting cost for past arrivals. On the other hand, processing the arrivals too soon leads to higher processing costs by missing the economies of scale of grouping larger numbers of arrivals into larger batches. Moreover, the timing of the next arrival is often unknown, meaning fixed-size batches or fixed waiting times tend to be poor choices. In this work, we consider the problem of finding the optimal batching schedule to minimize the sum of waiting time and processing cost under both offline and online settings. In the offline problem in which all arrival times are known a priori, we show that the optimal batching schedule can be found in polynomial time by reducing it to a shortest path problem on a weighted acyclic graph. For the online problem with unknown arrival times, we develop algorithms that are provably competitive for a broad range of processing-cost functions. We also provide a lower bound on the competitive ratio that no online algorithm can beat. Finally, we run numerical experiments on simulated and real data to demonstrate the effectiveness of our algorithms against the offline benchmark.

cs.DS↗

arXiv2025

A Theory of Inference Compute Scaling: Reasoning through Directed Stochastic Skill Search

Large language models (LLMs) demand considerable computational, energy, and financial resources during both training and deployment. While scaling laws for training have guided much of the field's recent progress, inference costs now represent a significant and growing component of the overall resource burden, particularly for reasoning-focused models. Existing characterizations of compute-optimality that consider model size, dataset size, and inference tokens in isolation or in fixed combinations risk overlooking more efficient operating points. We introduce directed stochastic skill search (DS3), a general framework that represents inference as stochastic traversal over a learned skill graph. From a simplified yet expressive instantiation, we derive closed-form expressions for task success and compute cost across a wide range of inference strategies -- including chain-of-thought (CoT) and tree-of-thought (ToT) -- enabling comparative analysis as a function of task difficulty and model capability. To that end, we extend a prior first-principles tripartite graph framework of LLM training to incorporate inference, and separately bridge DS3 with empirical methods that characterize LLM scaling behavior. We theoretically recover empirically observed patterns, including: linear accuracy scaling with logarithmic compute; variation in preferred inference strategies as a function of task difficulty and model capability; emergent behavior elicited by reasoning even when performance plateaus under parameter scaling; and both best-of-N (BoN) and majority voting behavior captured within a unified analytical framework. By explicitly characterizing training-inference interdependencies, our framework deepens theoretical understanding and supports principled algorithmic design and resource allocation.

cs.LG↗

arXiv2025

Spark: A System for Scientifically Creative Idea Generation

Recently, large language models (LLMs) have shown promising abilities to generate novel research ideas in science, a direction which coincides with many foundational principles in computational creativity (CC). In light of these developments, we present an idea generation system named Spark that couples retrieval-augmented idea generation using LLMs with a reviewer model named Judge trained on 600K scientific reviews from OpenReview. Our work is both a system demonstration and intended to inspire other CC researchers to explore grounding the generation and evaluation of scientific ideas within foundational CC principles. To this end, we release the annotated dataset used to train Judge, inviting other researchers to explore the use of LLMs for idea generation and creative evaluations.

cs.AI↗

arXiv2025

Transformational Creativity in Science: A Graphical Theory

Creative processes are typically divided into three types: combinatorial, exploratory, and transformational. Here, we provide a graphical theory of transformational scientific creativity, synthesizing Boden's insight that transformational creativity arises from changes in the "enabling constraints" of a conceptual space and Kuhn's structure of scientific revolutions as resulting from paradigm shifts. We prove that modifications made to axioms of our graphical model have the most transformative potential and then illustrate how several historical instances of transformational creativity can be captured by our framework.

cs.AI↗

arXiv2025

ITBench: Evaluating AI Agents across Diverse Real-World IT Automation Tasks

Realizing the vision of using AI agents to automate critical IT tasks depends on the ability to measure and understand effectiveness of proposed solutions. We introduce ITBench, a framework that offers a systematic methodology for benchmarking AI agents to address real-world IT automation tasks. Our initial release targets three key areas: Site Reliability Engineering (SRE), Compliance and Security Operations (CISO), and Financial Operations (FinOps). The design enables AI researchers to understand the challenges and opportunities of AI agents for IT automation with push-button workflows and interpretable metrics. ITBench includes an initial set of 94 real-world scenarios, which can be easily extended by community contributions. Our results show that agents powered by state-of-the-art models resolve only 13.8% of SRE scenarios, 25.2% of CISO scenarios, and 0% of FinOps scenarios. We expect ITBench to be a key enabler of AI-driven IT automation that is correct, safe, and fast.

cs.AI↗

arXiv2025

On Simultaneous Information and Energy Transmission through Quantum Channels

The optimal rate at which information can be sent through a quantum channel when the transmitted signal must simultaneously carry some minimum amount of energy is characterized. To do so, we introduce the quantum-classical analogue of the capacity-power function and generalize results in classical information theory for transmitting classical information through noisy channels. We show that the capacity-power function for a classical-quantum channel, for both unassisted and private protocol, is concave and also prove additivity for unentangled and uncorrelated ensembles of input signals for such channels. This implies we do not need regularized formulas for calculation. We show these properties also hold for all noiseless channels when we restrict the set of input states to be pure quantum states. For general channels, we find that the capacity-power function is piece-wise concave. We give an elegant visual proof for this supported by numerical simulations. We connect channel capacity and properties of random quantum states. In particular, we obtain analytical expressions for the capacity-power function for the case of noiseless channels using properties of random quantum states under an energy constraint and concentration phenomena in large Hilbert spaces.

quant-ph↗

arXiv2024

SwitchCIT: Switching for Continual Instruction Tuning

Large language models (LLMs) and multimodal models (MMs) have exhibited impressive capabilities in various domains, particularly in general language understanding and visual reasoning. However, these models, trained on massive data, may not be finely optimized for specific tasks triggered by instructions. Continual instruction tuning is crucial to adapt a large model to evolving tasks and domains, ensuring their effectiveness and relevance across a wide range of applications. In the context of continual instruction tuning, where models are sequentially trained on different tasks, catastrophic forgetting can occur, leading to performance degradation on previously learned tasks. This work addresses the catastrophic forgetting in continual instruction learning through a switching mechanism for routing computations to parameter-efficient tuned models. We demonstrate the effectiveness of our method through experiments on continual instruction tuning of different natural language generation tasks and vision-language tasks. We also showcase the advantages of our proposed method in terms of efficiency, scalability, portability, and privacy preservation.

cs.CL↗

arXiv2024

Semantically Grounded QFormer for Efficient Vision Language Understanding

General purpose Vision Language Models (VLMs) have received tremendous interest in recent years, owing to their ability to learn rich vision-language correlations as well as their broad zero-shot competencies. One immensely popular line of work utilizes frozen unimodal models, by bridging vision representations to language using a trainable module called the QFormer. However, this method relies heavily on large-scale multimodal pretraining with huge computational overheads. To that end, we propose a more efficient framework for QFormer-based vision-language alignment. Our key idea relies on the observation that QFormer latents correspond more strongly to the frozen LLM's intermediate latent space. Consequently, instead of using QFormer latents as inputs to the LLM, we alter the framework by using the latents to directly condition the LLM latent space for image-to-text generation. We demonstrate the effectiveness of our approach against existing baselines in improving the efficiency of vision-language pretraining.

cs.CV↗

arXiv2024

Suspense and surprise in the book of technology: Understanding innovation dynamics

We envision future technologies through science fiction, strategic planning, or academic research. Yet, our expectations do not always match with what actually unfolds, much like navigating a story where some events align with expectations while others surprise us. This gap indicates the inherent uncertainty of innovation-how technologies emerge and evolve in unpredictable ways. Here, we elaborate on this inherent uncertainty of innovation in the way technologies emerge and evolve. We define suspense captures accumulated uncertainty and describing events anticipated before their realization, while surprise represents a dramatic shift in understanding when an event occurs unexpectedly. We identify those connections in U.S. patents and show that suspenseful innovations tend to integrate more smoothly into society, achieving higher citations and market value. In contrast, surprising innovations, though often disruptive and groundbreaking, face challenges in adoption due to their extreme novelty. We further show that these categories allow us to identify distinct stages of technology life cycles, suggesting a way to identify the systematic trajectory of technologies and anticipate their future paths.

physics.soc-ph↗

arXiv2024

Formal Verification of Digital Twins with TLA and Information Leakage Control

Verifying the correctness of a digital twin provides a formal guarantee that the digital twin operates as intended. Digital twin verification is challenging due to the presence of uncertainties in the virtual representation, the physical environment, and the bidirectional flow of information between physical and virtual. A further challenge is that a digital twin of a complex system is composed of distributed components. This paper presents a methodology to specify and verify digital twin behavior, translating uncertain processes into a formally verifiable finite state machine. We use the Temporal Logic of Actions (TLA) to create a specification, an implementation abstraction that defines the properties required for correct system behavior. Our approach includes a novel weakening of formal security properties, allowing controlled information leakage while preserving theoretical guarantees. We demonstrate this approach on a digital twin of an unmanned aerial vehicle, verifying synchronization of physical-to-virtual and virtual-to-digital data flows to detect unintended misalignments.

cs.CR↗

arXiv2024

Compute-Update Federated Learning: A Lattice Coding Approach Over-the-Air

This paper introduces a federated learning framework that enables over-the-air computation via digital communications, using a new joint source-channel coding scheme. Without relying on channel state information at devices, this scheme employs lattice codes to both quantize model parameters and exploit interference from the devices. We propose a novel receiver structure at the server, designed to reliably decode an integer combination of the quantized model parameters as a lattice point for the purpose of aggregation. We present a mathematical approach to derive a convergence bound for the proposed scheme and offer design remarks. In this context, we suggest an aggregation metric and a corresponding algorithm to determine effective integer coefficients for the aggregation in each communication round. Our results illustrate that, regardless of channel dynamics and data heterogeneity, our scheme consistently delivers superior learning accuracy across various parameters and markedly surpasses other over-the-air methodologies.

cs.IT↗

arXiv2024

Online Reinforcement Learning with Passive Memory

This paper considers an online reinforcement learning algorithm that leverages pre-collected data (passive memory) from the environment for online interaction. We show that using passive memory improves performance and further provide theoretical guarantees for regret that turns out to be near-minimax optimal. Results show that the quality of passive memory determines sub-optimality of the incurred regret. The proposed approach and results hold in both continuous and discrete state-action spaces.

cs.LG↗

arXiv2024

From Phytochemicals to Recipes: Health Indications and Culinary Uses of Herbs and Spices

Herbs and spices each contain about 3000 phytochemicals on average and there is much traditional knowledge on their health benefits. However, there is a lack of systematic study to understand the relationship among herbs and spices, their phytochemical constituents, their potential health benefits, and their usage in regional cuisines. Here we use a network-based approach to elucidate established relationships and predict novel associations between the phytochemicals present in herbs and spices with health indications. Our top 100 inferred indication-phytochemical relationships rediscover 40% known relationships and 20% that have been inferred via gene-chemical interactions with high confidence. The remaining 40% are hypotheses generated in a principled way for further experimental investigations. We also develop an algorithm to find the minimum set of spices needed to cover a target group of health conditions. Drawing on spice usage patterns in several regional Indian cuisines, and a copy-mutate model for regional cuisine evolution, we characterize the spectrum of health conditions covered by existing regional cuisines. The spectrum of health conditions can expand through the nationalization/globalization of culinary practice.

physics.soc-ph↗

arXiv2024

An Information Theory of Compute-Optimal Size Scaling, Emergence, and Plateaus in Language Models

Recent empirical studies show three phenomena with increasing size of language models: compute-optimal size scaling, emergent capabilities, and performance plateauing. We present a simple unified mathematical framework to explain all of these language model scaling phenomena, building on recent skill-text bipartite graph frameworks for semantic learning. Modeling the learning of concepts from texts as an iterative process yields an analogy to iterative decoding of low-density parity check (LDPC) codes in information theory. Thence, drawing on finite-size scaling characterizations of LDPC decoding, we derive the compute-optimal size scaling (Chinchilla rule) for language models. Further, using tools from random network theory, we provide a simple explanation for both emergence of complex skills and plateauing of performance as the size of language models scale. We see multiple plateaus.

cs.IT↗

arXiv2024

Efficient Model-Agnostic Multi-Group Equivariant Networks

Constructing model-agnostic group equivariant networks, such as equitune (Basu et al., 2023b) and its generalizations (Kim et al., 2023), can be computationally expensive for large product groups. We address this problem by providing efficient model-agnostic equivariant designs for two related problems: one where the network has multiple inputs each with potentially different groups acting on them, and another where there is a single input but the group acting on it is a large product group. For the first design, we initially consider a linear model and characterize the entire equivariant space that satisfies this constraint. This characterization gives rise to a novel fusion layer between different channels that satisfies an invariance-symmetry (IS) constraint, which we call an IS layer. We then extend this design beyond linear models, similar to equitune, consisting of equivariant and IS layers. We also show that the IS layer is a universal approximator of invariant-symmetric functions. Inspired by the first design, we use the notion of the IS property to design a second efficient model-agnostic equivariant design for large product groups acting on a single input. For the first design, we provide experiments on multi-image classification where each view is transformed independently with transformations such as rotations. We find equivariant models are robust to such transformations and perform competitively otherwise. For the second design, we consider three applications: language compositionality on the SCAN dataset to product groups; fairness in natural language generation from GPT-2 to address intersectionality; and robust zero-shot image classification with CLIP. Overall, our methods are simple and general, competitive with equitune and its variants, while also being computationally more efficient.

cs.LG↗

arXiv2024

Persona Inconstancy in Multi-Agent LLM Collaboration: Conformity, Confabulation, and Impersonation

Multi-agent AI systems can be used for simulating collective decision-making in scientific and practical applications. They can also be used to introduce a diverse group discussion step in chatbot pipelines, enhancing the cultural sensitivity of the chatbot's responses. These applications, however, are predicated on the ability of AI agents to reliably adopt assigned personas and mimic human interactions. To see whether LLM agents satisfy these requirements, we examine AI agent ensembles engaged in cross-national collaboration and debate by analyzing their private responses and chat transcripts. Our findings suggest that multi-agent discussions can support collective AI decisions that more often reflect diverse perspectives, yet this effect is tempered by the agents' susceptibility to conformity due to perceived peer pressure and occasional challenges in maintaining consistent personas and opinions. Instructions that encourage debate in support of one's opinions rather than collaboration increase the rate of inconstancy. Without addressing the factors we identify, the full potential of multi-agent frameworks for producing more culturally diverse AI outputs or more realistic simulations of group decision-making may remain untapped.

cs.AI↗

arXiv2024

Fractional Budget Allocation for Influence Maximization under General Marketing Strategies

We consider the fractional influence maximization problem, i.e., identifying users on a social network to be incentivized with potentially partial discounts to maximize the influence on the network. The larger the discount given to a user, the higher the likelihood of its activation (adopting a new product or innovation), who then attempts to activate its neighboring users, causing a cascade effect of influence through the network. Our goal is to devise efficient algorithms that assign initial discounts to the network's users to maximize the total number of activated users at the end of the cascade, subject to a constraint on the total sum of discounts given. In general, the activation likelihood could be any non-decreasing function of the discount, whereas, our focus lies on the case when the activation likelihood is an affine function of the discount, potentially varying across different users. As this problem is shown to be NP-hard, we propose and analyze an efficient (1-1/e)-approximation algorithm. Furthermore, we run experiments on real-world social networks to show the performance and scalability of our method.

cs.SI↗

arXiv2024

Reliable Quantum Memories with Unreliable Components

Quantum memory systems are vital in quantum information processing for dependable storage and retrieval of quantum states. Inspired by classical reliability theories that synthesize reliable computing systems from unreliable components, we formalize the problem of reliable storage of quantum information using noisy components. We introduce the notion of stable quantum memories and define the storage rate as the ratio of the number of logical qubits to the total number of physical qubits, as well as the circuit complexity of the decoder, which includes both quantum gates and measurements. We demonstrate that a strictly positive storage rate can be achieved by constructing a quantum memory system with quantum expander codes. Moreover, by reducing the reliable storage problem to reliable quantum communication, we provide upper bounds on the achievable storage capacity. In the case of physical qubits corrupted by noise satisfying hypercontractivity conditions, we provide a tighter upper bound on storage capacity using an entropy dissipation argument. Furthermore, observing that the time complexity of the decoder scales non-trivially with the number of physical qubits, achieving asymptotic rates may not be possible due to the induced dependence of the noise on the number of physical qubits. In this constrained non-asymptotic setting, we derive upper bounds on storage capacity using finite blocklength communication bounds. Finally, we numerically analyze the gap between upper and lower bounds in both asymptotic and non-asymptotic cases, and provide suggestions to tighten the gap.

quant-ph↗