Searcharxiv⌕ Search

arXiv subjects

Ankit Gupta

Publications and source records attributed to Ankit Gupta.

At least 37 records · Page 2Linked to original sources

Data Guard: A Fine-grained Purpose-based Access Control System for Large Data Warehouses

The last few years have witnessed a spate of data protection regulations in conjunction with an ever-growing appetite for data usage in large businesses, which presents significant challenges for businesses to maintain compliance. To address this conflict, we present Data Guard - a fine-grained, purpose-based access control system for large data warehouses. Data Guard enables authoring policies based on semantic descriptions of data and purpose of data access. Data Guard then translates these policies into SQL views that mask data from the underlying warehouse tables. At access time, Data Guard ensures compliance by transparently routing each table access to the appropriate data-masking view based on the purpose of the access, thus minimizing the effort of adopting Data Guard in existing applications. Our enforcement solution allows masking data at much finer granularities than what traditional solutions allow. In addition to row and column level data masking, Data Guard can mask data at the sub-cell level for columns with non-atomic data types such as structs, arrays, and maps. This fine-grained masking allows Data Guard to preserve data utility for consumers while ensuring compliance. We implemented a number of performance optimizations to minimize the overhead of data masking operations. We perform numerous experiments to identify the key factors that influence the data masking overhead and demonstrate the efficiency of our implementation. Data Guard is deployed inside LinkedIn's production data warehouses and ensures compliance of more than 20,000 table accesses each day across different data processing engines.

cs.DB↗

Entropic active particle transport in pulsating 3D geometries

We study the transport of active Brownian particles (ABPs) in three-dimensional (3D) oscillatory geometries, which are spatially periodic. We establish a generalized Fick-Jacobs approach, which reduces a 3D system to an effective 1D system based on the assumption that a fast equilibration of particles along the transversal directions of the geometry. The transport characteristics of ABPs are computed semi-analytically and corroborated by numerical simulations. At the optimal frequency of the geometry oscillation, particles exhibit higher average velocity $\langle v \rangle$ and effective diffusion coefficient $D_{\text{eff}}$, resembling the phenomena of stochastic resonance. This effect is further enhanced by the self-propelled velocity of ABPs and the amplitude of geometry oscillations. These findings have significant implications for the development of micro- and nanofluidic devices with enhanced control over particle transport and precise manipulation of small-scale biomedical devices.

cond-mat.soft↗

Benchmarking and Evaluation of AI Models in Biology: Outcomes and Recommendations from the CZI Virtual Cells Workshop

Artificial intelligence holds immense promise for transforming biology, yet a lack of standardized, cross domain, benchmarks undermines our ability to build robust, trustworthy models. Here, we present insights from a recent workshop that convened machine learning and computational biology experts across imaging, transcriptomics, proteomics, and genomics to tackle this gap. We identify major technical and systemic bottlenecks such as data heterogeneity and noise, reproducibility challenges, biases, and the fragmented ecosystem of publicly available resources and propose a set of recommendations for building benchmarking frameworks that can efficiently compare ML models of biological systems across tasks and data modalities. By promoting high quality data curation, standardized tooling, comprehensive evaluation metrics, and open, collaborative platforms, we aim to accelerate the development of robust benchmarks for AI driven Virtual Cells. These benchmarks are crucial for ensuring rigor, reproducibility, and biological relevance, and will ultimately advance the field toward integrated models that drive new discoveries, therapeutic insights, and a deeper understanding of cellular systems.

cs.LG↗

Role of obstacle softness in the diffusive behavior of active Particles

We numerically investigate the diffusive behavior of active Brownian particles in a two-dimensional confined channel filled with soft obstacles, whose softness is controlled by a parameter $K$. Here, active particles are subjected to external bias $F$. Particle diffusion is influenced by entropic barriers that arise due to variations in the shape of the chosen channel geometry. We observed that the interplay between obstacle softness, entropic barriers, and external bias leads to striking transport characteristics of the active particles. For instance, with increasing $F$, the non-linear mobility exhibits non-monotonic behavior, and effective diffusion is greatly enhanced, showing multiple peaks in the presence of soft obstacles. Further, as a function of $K$ and $F$, particles exhibit various diffusive behaviors, e.g., normal diffusion - where the role of obstacles is insignificant, subdiffusion or superdiffusion - where the particles are partially trapped by the obstacles, and particles are ultimately caged by the obstacles. These findings help understand the physical situations wherein active agents diffuse in crowded environments.

cond-mat.soft↗

Unbiased estimation of second-order parameter sensitivities for stochastic reaction networks

Stochastic models for chemical reaction networks are increasingly popular in systems and synthetic biology. These models formulate the reaction dynamics as Continuous-Time Markov Chains (CTMCs) whose propensities are parameterized by a vector $θ$ and parameter sensitivities are introduced as derivatives of their expected outputs with respect to components of the parameter vector. Sensitivities characterise key properties of the output like robustness and are also at the heart of numerically efficient optimisation routines like Newton-type algorithms used in parameter inference and the design of of control mechanisms. Currently the only unbiased estimator for second-order sensitivities is based on the Girsanov transform and it often suffers from high estimator variance. We develop a novel estimator for second-order sensitivities by first rigorously deriving an integral representation of these sensitivities. We call the resulting method the Double Bernoulli Path Algorithm and illustrate its efficiency through numerical examples.

q-bio.MN↗

How to Build the Virtual Cell with Artificial Intelligence: Priorities and Opportunities

The cell is arguably the most fundamental unit of life and is central to understanding biology. Accurate modeling of cells is important for this understanding as well as for determining the root causes of disease. Recent advances in artificial intelligence (AI), combined with the ability to generate large-scale experimental data, present novel opportunities to model cells. Here we propose a vision of leveraging advances in AI to construct virtual cells, high-fidelity simulations of cells and cellular systems under different conditions that are directly learned from biological data across measurements and scales. We discuss desired capabilities of such AI Virtual Cells, including generating universal representations of biological entities across scales, and facilitating interpretable in silico experiments to predict and understand their behavior using virtual instruments. We further address the challenges, opportunities and requirements to realize this vision including data needs, evaluation strategies, and community standards and engagement to ensure biological accuracy and broad utility. We envision a future where AI Virtual Cells help identify new drug targets, predict cellular responses to perturbations, as well as scale hypothesis exploration. With open science collaborations across the biomedical ecosystem that includes academia, philanthropy, and the biopharma and AI industries, a comprehensive predictive understanding of cell mechanisms and interactions has come into reach.

q-bio.QM↗

SpikingRx: From Neural to Spiking Receiver

In this work, we propose an energy efficient neuromorphic receiver to replace multiple signal-processing blocks at the receiver by a Spiking Neural Network (SNN) based module, called SpikingRx. We propose a deep convolutional SNN with spike-element-wise ResNet layers which takes a whole OFDM grid compliant with 5G specifications and provides soft outputs for decoded bits that can be used as log-likelihood ratios. We propose to employ the surrogate gradient descent method for training the SpikingRx and focus on its generalizability and robustness to quantization. Moreover, the interpretability of the proposed SpikingRx is studied by a comprehensive ablation study. Our extensive numerical simulations show that SpikingRx is capable of achieving significant block error rate performance gain compared to conventional 5G receivers and similar performance compared to its traditional NN-based counterparts with approximately 9x less energy consumption.

cs.IT↗

The generator gradient estimator is an adjoint state method for stochastic differential equations

Motivated by the increasing popularity of overparameterized Stochastic Differential Equations (SDEs) like Neural SDEs, Wang, Blanchet and Glynn recently introduced the generator gradient estimator, a novel unbiased stochastic gradient estimator for SDEs whose computation time remains stable in the number of parameters. In this note, we demonstrate that this estimator is in fact an adjoint state method, an approach which is known to scale with the number of states and not the number of parameters in the case of Ordinary Differential Equations (ODEs). In addition, we show that the generator gradient estimator is a close analogue to the exact Integral Path Algorithm (eIPA) estimator which was introduced by Gupta, Rathinam and Khammash for a class of Continuous-Time Markov Chains (CTMCs) known as stochastic chemical reactions networks (CRNs).

math.OC↗

Never Train from Scratch: Fair Comparison of Long-Sequence Models Requires Data-Driven Priors

Modeling long-range dependencies across sequences is a longstanding goal in machine learning and has led to architectures, such as state space models, that dramatically outperform Transformers on long sequences. However, these impressive empirical gains have been by and large demonstrated on benchmarks (e.g. Long Range Arena), where models are randomly initialized and trained to predict a target label from an input sequence. In this work, we show that random initialization leads to gross overestimation of the differences between architectures and that pretraining with standard denoising objectives, using $\textit{only the downstream task data}$, leads to dramatic gains across multiple architectures and to very small gaps between Transformers and state space models (SSMs). In stark contrast to prior works, we find vanilla Transformers to match the performance of S4 on Long Range Arena when properly pretrained, and we improve the best reported results of SSMs on the PathX-256 task by 20 absolute points. Subsequently, we analyze the utility of previously-proposed structured parameterizations for SSMs and show they become mostly redundant in the presence of data-driven initialization obtained through pretraining. Our work shows that, when evaluating different architectures on supervised tasks, incorporation of data-driven priors via pretraining is essential for reliable performance estimation, and can be done efficiently.

cs.LG↗

Exploring the limits of decoder-only models trained on public speech recognition corpora

The emergence of industrial-scale speech recognition (ASR) models such as Whisper and USM, trained on 1M hours of weakly labelled and 12M hours of audio only proprietary data respectively, has led to a stronger need for large scale public ASR corpora and competitive open source pipelines. Unlike the said models, large language models are typically based on Transformer decoders, and it remains unclear if decoder-only models trained on public data alone can deliver competitive performance. In this work, we investigate factors such as choice of training datasets and modeling components necessary for obtaining the best performance using public English ASR corpora alone. Our Decoder-Only Transformer for ASR (DOTA) model comprehensively outperforms the encoder-decoder open source replication of Whisper (OWSM) on nearly all English ASR benchmarks and outperforms Whisper large-v3 on 7 out of 15 test sets. We release our codebase and model checkpoints under permissive license.

cs.CL↗

Analyzing Transformers in Embedding Space

Understanding Transformer-based models has attracted significant attention, as they lie at the heart of recent technological advances across machine learning. While most interpretability methods rely on running models over inputs, recent work has shown that a zero-pass approach, where parameters are interpreted directly without a forward/backward pass is feasible for some Transformer parameters, and for two-layer attention networks. In this work, we present a theoretical analysis where all parameters of a trained Transformer are interpreted by projecting them into the embedding space, that is, the space of vocabulary items they operate on. We derive a simple theoretical framework to support our arguments and provide ample evidence for its validity. First, an empirical analysis showing that parameters of both pretrained and fine-tuned models can be interpreted in embedding space. Second, we present two applications of our framework: (a) aligning the parameters of different models that share a vocabulary, and (b) constructing a classifier without training by ``translating'' the parameters of a fine-tuned classifier to parameters of a different model that was only pretrained. Overall, our findings open the door to interpretation methods that, at least in part, abstract away from model specifics and operate in the embedding space only.

cs.CL↗

Simplifying and Understanding State Space Models with Diagonal Linear RNNs

Sequence models based on linear state spaces (SSMs) have recently emerged as a promising choice of architecture for modeling long range dependencies across various modalities. However, they invariably rely on discretization of a continuous state space, which complicates their presentation and understanding. In this work, we dispose of the discretization step, and propose a model based on vanilla Diagonal Linear RNNs ($\mathrm{DLR}$). We empirically show that, despite being conceptually much simpler, $\mathrm{DLR}$ is as performant as previously-proposed SSMs on a variety of tasks and benchmarks including Long Range Arena and raw speech classification. Moreover, we characterize the expressivity of SSMs (including $\mathrm{DLR}$) and attention-based models via a suite of $13$ synthetic sequence-to-sequence tasks involving interactions over tens of thousands of tokens, ranging from simple operations, such as shifting an input sequence, to detecting co-dependent visual features over long spatial ranges in flattened images. We find that while SSMs report near-perfect performance on tasks that can be modeled via $\textit{few}$ convolutional kernels, they struggle on tasks requiring $\textit{many}$ such kernels and especially when the desired sequence manipulation is $\textit{context-dependent}$. Despite these limitations, $\mathrm{DLR}$ reaches high performance on two higher-order reasoning tasks $\mathrm{ListOpsSubTrees}$ and $\mathrm{PathfinderSegmentation}\text{-}\mathrm{256}$ with input lengths $8K$ and $65K$ respectively, and gives encouraging performance on $\mathrm{PathfinderSegmentation}\text{-}\mathrm{512}$ with input length $262K$ for which attention is not a viable choice.

cs.LG↗

Data-Driven Computation of Robust Invariant Sets and Gain-Scheduled Controllers for Linear Parameter-Varying Systems

We present a direct data-driven approach to synthesize robust control invariant (RCI) sets and their associated gain-scheduled feedback control laws for linear parameter-varying (LPV) systems subjected to bounded disturbances. A data-set consisting of a single state-input-scheduling trajectory is gathered from the system, which is directly utilized to compute polytopic RCI set and controllers by solving a semidefinite program. The proposed method does not require an intermediate LPV model identification step. Through a numerical example, we show that the proposed approach can generate RCI sets with a relatively small number of data samples when the data satisfies certain excitation conditions.

eess.SY↗

Effective filtering approach for joint parameter-state estimation in SDEs via Rao-Blackwellization and modularization

Stochastic filtering is a vibrant area of research in both control theory and statistics, with broad applications in many scientific fields. Despite its extensive historical development, there still lacks an effective method for joint parameter-state estimation in SDEs. The state-of-the-art particle filtering methods suffer from either sample degeneracy or information loss, with both issues stemming from the dynamics of the particles generated to represent system parameters. This paper provides a novel and effective approach for joint parameter-state estimation in SDEs via Rao-Blackwellization and modularization. Our method operates in two layers: the first layer estimates the system states using a bootstrap particle filter, and the second layer marginalizes out system parameters explicitly. This strategy circumvents the need to generate particles representing system parameters, thereby mitigating their associated problems of sample degeneracy and information loss. Moreover, our method employs a modularization approach when integrating out the parameters, which significantly reduces the computational complexity. All these designs ensure the superior performance of our method. Finally, a numerical example is presented to illustrate that our method outperforms existing approaches by a large margin.

math.OC↗

Efficient High-Resolution Template Matching with Vector Quantized Nearest Neighbour Fields

Template matching is a fundamental problem in computer vision with applications in fields including object detection, image registration, and object tracking. Current methods rely on nearest-neighbour (NN) matching, where the query feature space is converted to NN space by representing each query pixel with its NN in the template. NN-based methods have been shown to perform better in occlusions, appearance changes, and non-rigid transformations; however, they scale poorly with high-resolution data and high feature dimensions. We present an NN-based method which efficiently reduces the NN computations and introduces filtering in the NN fields (NNFs). A vector quantization step is introduced before the NN calculation to represent the template with $k$ features, and the filter response over the NNFs is used to compare the template and query distributions over the features. We show that state-of-the-art performance is achieved in low-resolution data, and our method outperforms previous methods at higher resolution.

cs.CV↗

Direct Data-Driven Computation of Polytopic Robust Control Invariant Sets and State-Feedback Controllers

This paper presents a direct data-driven approach for computing robust control invariant (RCI) sets and their associated state-feedback control laws for linear time-invariant systems affected by bounded disturbances. The proposed method utilizes a single state-input trajectory generated from the system, to compute a polytopic RCI set with a desired complexity and an invariance-inducing feedback controller, without the need to identify a model of the system. The problem is formulated in terms of a set of sufficient linear matrix inequality conditions that are then combined in a semi-definite program to maximize the volume of the RCI set while respecting the state and input constraints. We demonstrate through a numerical case study that the proposed data-driven approach can generate RCI sets that are of comparable size to those obtained by a model-based method in which exact knowledge of the system matrices is assumed.

eess.SY↗

Separation of interacting active particles in an asymmetric channel

We study the diffusive behaviour of interacting active particles (self-propelled) with mass $m$ in an asymmetric channel. The particles are subjected to an external oscillatory force along the length of the channel. In this setup, particles may exhibit rectification. In the absence of interaction, the mean velocity $\langle v \rangle$ of the particles shows a maximum at moderate $m$ values. It means that particles of moderate $m$ have higher velocities than the others. However, by incorporating short-range interaction between the particles, $\langle v \rangle$ exhibits an additional peak at lower $m$ values, indicating that particles of lower and moderate m can be separated simultaneously from the rest. Furthermore, by tuning the interaction strength, the self-propelled velocity, and the parameters of the oscillatory force, one can selectively separate the particles of lower $m$, moderate $m$, or both. Empirical relations for estimating the optimal mass as a function of these parameters are discussed. These findings are beneficial for separating the particles of selective $m$ from the rest of the particles.

cond-mat.soft↗

Complete characterization of robust perfect adaptation in biochemical reaction networks

Perfect adaptation is a phenomenon whereby the output variables of a system can maintain certain values despite external disturbances. Robust perfect adaptation (RPA) refers to an adaptation property that does not require fine-tuning of system parameters. RPA plays a vital role for the survival of living systems in unpredictable environments. However, complex interaction patterns in biochemical systems pose a significant challenge in identifying RPA and associated regulatory mechanisms. The goal of this paper is to present a novel approach for finding all RPA properties that are realized for a generic choice of kinetics for general deterministic chemical reaction systems. This is accomplished by proving that an RPA property is represented by a subnetwork with certain topological features. This connection is exploited to show that these structures generate all kinetics-independent RPA properties, allowing us to systematically identify all RPA properties by enumerating these subnetworks. An efficient method is developed for this enumeration, and we provide a computational package for this purpose. We pinpoint the integral feedback controllers that work in concert to realize each RPA property, casting our results into the familiar control-theoretic paradigm of the Internal Model Principle. We further generalize the regulation problem to the multi-output scenario where the target values belong to a manifold of nonzero dimension, and provide a sufficient condition for this. The present work significantly advances our understanding of regulatory mechanisms that lead to RPA in endogenous biochemical systems, and it also provides rational design principles for synthetic controllers. The present results indicate that an RPA property is essentially equivalent to the existence of a "topological invariant", which is an instance of what we call the "Robust Adaptation is Topological"(RAT) principle.

q-bio.MN↗