arXiv subjects

Saumya Gupta

Publications and source records attributed to Saumya Gupta.

At least 19 recordsLinked to original sources

arXiv2026

Revealing the substellar population of IC 1396: A spectroscopic survey of brown dwarfs in the region

We present a new spectroscopic view of the brown dwarf population in the young star-forming region IC 1396 and investigate the impact of environment on low-mass star formation. We use deep optical photometry from Subaru-HSC to identify the candidate low-mass stars and brown dwarfs in the region. Our follow-up low-resolution spectroscopic survey with GTC-EMIR and IRTF-SpeX has identified 32 new members in the region with spectral types between M3 and M9, among which 25 are brown dwarfs with spectral types M6 or later. We use the BT-Settl atmospheric models to derive the effective temperatures of the members. Using a comprehensive catalogue of known members and candidates, we estimate the star to brown dwarf ratio for IC 1396 to be 5.0$\pm$0.4, for a mass range between 1-0.03 M$_\odot$. This ratio is largely consistent with measurements in other young clusters spanning a range of UV radiation fields and stellar densities, supporting formation scenarios in which the relative abundance of brown dwarfs is not strongly influenced by the local environmental conditions.

astro-ph.SR↗

arXiv2026

DeepWeightFlow: Re-Basined Flow Matching for Generating Neural Network Weights

Building efficient and effective generative models for neural network weights has been a research focus of significant interest that faces challenges posed by the high-dimensional weight spaces of modern neural networks and their symmetries. Several prior generative models are limited to generating partial neural network weights, particularly for larger models, such as ResNet and ViT. Those that do generate complete weights struggle with generation speed or require finetuning of the generated models. In this work, we present DeepWeightFlow, a Flow Matching model that operates directly in weight space to generate diverse and high-accuracy neural network weights for a variety of architectures, neural network sizes, and data modalities. The neural networks generated by DeepWeightFlow do not require fine-tuning to perform well and can scale to large networks. We apply Git Re-Basin and TransFusion for neural network canonicalization in the context of generative weight models to account for the impact of neural network permutation symmetries and to improve generation efficiency for larger model sizes. The generated networks excel at transfer learning, and ensembles of hundreds of neural networks can be generated in minutes, far exceeding the efficiency of diffusion-based methods. DeepWeightFlow models pave the way for more efficient and scalable generation of diverse sets of neural networks.

cs.LG↗

arXiv2026

Image Synthesis Using Spintronic Deep Convolutional Generative Adversarial Network

The computational requirements of generative adversarial networks (GANs) exceed the limit of conventional Von Neumann architectures, necessitating energy efficient alternatives such as neuromorphic spintronics. This work presents a hybrid CMOS-spintronic deep convolutional generative adversarial network (DCGAN) architecture for synthetic image generation. The proposed generative vision model approach follows the standard framework, leveraging generator and discriminators adversarial training with our designed spintronics hardware for deconvolution, convolution, and activation layers of the DCGAN architecture. To enable hardware aware spintronic implementation, the generator's deconvolution layers are restructured as zero padded convolution, allowing seamless integration with a 6-bit skyrmion based synapse in a crossbar, without compromising training performance. Nonlinear activation functions are implemented using a hybrid CMOS domain wall based Rectified linear unit (ReLU) and Leaky ReLU units. Our proposed tunable Leaky ReLU employs domain wall position coded, continuous resistance states and a piecewise uniaxial parabolic anisotropy profile with a parallel MTJ readout, exhibiting energy consumption of 0.192 pJ. Our spintronic DCGAN model demonstrates adaptability across both grayscale and colored datasets, achieving Fr'echet Inception Distances (FID) of 27.5 for the Fashion MNIST and 45.4 for Anime Face datasets, with testing energy (training energy) of 4.9 nJ (14.97~nJ/image) and 24.72 nJ (74.7 nJ/image).

physics.app-ph↗

arXiv2025

LoC-Path: Learning to Compress for Pathology Multimodal Large Language Models

Whole Slide Image (WSI) MLLMs are difficult to build and deploy because gigapixel slides induce thousands of visual tokens, while only a small fraction of regions is diagnostically relevant. Existing slide-level pathology MLLMs typically combine heavy slide-level encoders with long visual prefixes, making end-to-end slide-level development and deployment expensive under limited computational resources. We revisit this regime and show that WSI tile features are highly redundant at both global and local scales, while task-relevant evidence is sparse and query-dependent. We therefore introduce LoC-Path, a resource-efficient slide-level MLLM that compresses before fusion. LoC-Path uses a Sparse Token Merger (STM) and an MAE-pretrained resampler to replace expensive slide-level encoding with a compact latent interface, then uses a Token Importance Scorer (TIS) to select the most relevant latents and a Cross-Attention Routing Adapter (CARA) to fuse them into a few LLM decoder layers. This design lowers both multimodal tuning cost and inference-time latency/memory by avoiding heavy slide-level encoding and long visual prefixes. Extensive experiments show that LoC-Path remains competitive with prior slide-level MLLMs while making end-to-end development and deployment more practical under limited computational resources.

cs.CV↗

arXiv2025

FLAT: FLow-Aligned Training of Unrolled Networks for MRI Reconstruction

Unrolled networks are widely used in Magnetic Resonance Imaging (MRI) reconstruction for their efficiency. Structured as a series of neural network stages (or cascades), an unrolled network takes a low-quality input and passes it sequentially through each stage to iteratively refine the reconstruction. However, unrolled networks typically exhibit unstable output quality across cascades, resulting in sub-optimal final reconstruction results. In this work, we address this inherent limitation of unrolled networks, drawing inspiration from recent Flow Matching paradigm. We first theoretically show that unrolled networks can be viewed as discretizations of approximate conditional probability flows. This connection shows that unrolled networks and Flow Matching are analogous in MRI reconstruction. Building upon this insight, we propose FLow-Aligned Training (FLAT), which (1) derives important cascade parameters from the Flow Matching discretization; and (2) aligns intermediate reconstructions with the ideal Flow Matching trajectory to improve cascade iteration stability and convergence. Experiments on three MRI datasets show that FLAT results in a stable trajectory across sub-networks, improving the quality of the final reconstruction.

cs.CV↗

arXiv2025

Bridging 3D Deep Learning and Curation for Analysis and High-Quality Segmentation in Practice

Accurate 3D microscopy image segmentation is critical for quantitative bioimage analysis but even state-of-the-art foundation models yield error-prone results. Therefore, manual curation is still widely used for either preparing high-quality training data or fixing errors before analysis. We present VessQC, an open-source tool for uncertainty-guided curation of large 3D microscopy segmentations. By integrating uncertainty maps, VessQC directs user attention to regions most likely containing biologically meaningful errors. In a preliminary user study uncertainty-guided correction significantly improved error detection recall from 67% to 94.0% (p=0.007) without a significant increase in total curation time. VessQC thus enables efficient, human-in-the-loop refinement of volumetric segmentations and bridges a key gap in real-world applications between uncertainty estimation and practical human-computer interaction. The software is freely available at github.com/MMV-Lab/VessQC.

cs.CV↗

arXiv2025

Efficient Whole Slide Pathology VQA via Token Compression

Whole-slide images (WSIs) in pathology can reach up to 10,000 x 10,000 pixels, posing significant challenges for multimodal large language model (MLLM) due to long context length and high computational demands. Previous methods typically focus on patch-level analysis or slide-level classification using CLIP-based models with multi-instance learning, but they lack the generative capabilities needed for visual question answering (VQA). More recent MLLM-based approaches address VQA by feeding thousands of patch tokens directly into the language model, which leads to excessive resource consumption. To address these limitations, we propose Token Compression Pathology LLaVA (TCP-LLaVA), the first MLLM architecture to perform WSI VQA via token compression. TCP-LLaVA introduces a set of trainable compression tokens that aggregate visual and textual information through a modality compression module, inspired by the [CLS] token mechanism in BERT. Only the compressed tokens are forwarded to the LLM for answer generation, significantly reducing input length and computational cost. Experiments on ten TCGA tumor subtypes show that TCP-LLaVA outperforms existing MLLM baselines in VQA accuracy while reducing training resource consumption by a substantial margin.

cs.CV↗

arXiv2025

BrainMRDiff: A Diffusion Model for Anatomically Consistent Brain MRI Synthesis

Accurate brain tumor diagnosis relies on the assessment of multiple Magnetic Resonance Imaging (MRI) sequences. However, in clinical practice, the acquisition of certain sequences may be affected by factors like motion artifacts or contrast agent contraindications, leading to suboptimal outcome, such as poor image quality. This can then affect image interpretation by radiologists. Synthesizing high quality MRI sequences has thus become a critical research focus. Though recent advancements in controllable generative AI have facilitated the synthesis of diagnostic quality MRI, ensuring anatomical accuracy remains a significant challenge. Preserving critical structural relationships between different anatomical regions is essential, as even minor structural or topological inconsistencies can compromise diagnostic validity. In this work, we propose BrainMRDiff, a novel topology-preserving, anatomy-guided diffusion model for synthesizing brain MRI, leveraging brain and tumor anatomies as conditioning inputs. To achieve this, we introduce two key modules: Tumor+Structure Aggregation (TSA) and Topology-Guided Anatomy Preservation (TGAP). TSA integrates diverse anatomical structures with tumor information, forming a comprehensive conditioning mechanism for the diffusion process. TGAP enforces topological consistency during reverse denoising diffusion process; both these modules ensure that the generated image respects anatomical integrity. Experimental results demonstrate that BrainMRDiff surpasses existing baselines, achieving performance improvements of 23.33% on the BraTS-AG dataset and 33.33% on the BraTS-Met dataset. Code will be made publicly available soon.

eess.IV↗

arXiv2025

Subaru Hyper-Supreme Cam observations of IC 1396: Source catalogue, member population, and sub-clusters of the complex

To identify member populations of IC 1396, we employ the random forest (RF) classifier of machine learning technique. Random forest classifier is an ensemble of individual decision trees suitable for large, high-dimensional datasets. The training set used in this work is derived from previous Gaia-based studies, where the member stars are younger than $\sim$ 10~Myr. However, its sensitivity is limited to $\sim$ 20~mag in the $\rm r_2$ band, making it challenging to identify candidates at the fainter end. In this analysis, in addition to magnitudes and colours, we incorporate several derived parameters from the magnitude and colour of the sources to identify candidate members of the star-forming complex. By employing this method, we are able to identify promising candidate member populations of the star-forming complex. We discuss the associated limitations and caveats in the method and for improvment in future studies. In this analysis, we identify 2425 high-probability low-mass stars distributed within the entire star-forming complex, of which 1331 are new detections. Comparison of these identified member populations shows a high retrieval rate with Gaia-based literature sources, as well as sources detected through methods based on optical spectroscopy, Spitzer, $\rm H_{\alpha}/X-ray$ emissions, optical, and 2MASS photometry. The mean age of the member populations is $\rm \sim 2-4~Myr$, consistent with findings from previous studies. Considering the identified member populations, we present preliminary results by exploring the presence of sub-clusters within IC 1396, assessing the possible mass limit of the member populations, and providing a brief discussion on the star formation history of the complex.

astro-ph.GA↗

arXiv2025

The past, present and future of observations of externally irradiated disks

Recent years have seen a surge of interest in the community studying the effect of ultraviolet radiation environment, predominantly set by OB stars, on protoplanetary disc evolution and planet formation. This is important because a significant fraction of planetary systems, potentially including our own, formed in close proximity to OB stars. This is a rapidly developing field, with a broad range of observations across many regions recently obtained or recently scheduled. In this paper, stimulated by a series of workshops on the topic, we take stock of the current and upcoming observations. We discuss how the community can build on this recent success with future observations to make progress in answering the big questions of the field, with the broad goal of disentangling how external photoevaporation contributes to shaping the observed (exo)planet population. Both existing and future instruments offer numerous opportunities to make progress towards this goal.

astro-ph.SR↗

arXiv2024

Gummy's Way Out -- a Tangible Interactive Narrative with Food and the Diegetic Body

There is growing interest in designing playful interactions with food, but food based tangible interactive narratives have received less attention. We introduce Gummy's Way Out, an interactive tangible narrative experience where interactors eat a gummy bear and help him find his way out of their bodies by eating various food items. By consuming different things, the interactor either helps or hinders the gummy bear's journey through an imagined Diegetic body that overlaps with their own. Interactors are endowed with the gummy bear's well-being and are also encouraged to reflect on how their actions can impact their Lived body. We present preliminary results of a user study and design considerations on how to design for the diegetic body in interactive food based narrative experiences. We recommend leveraging the sensory and emotional properties of food to create a visceral narrative experience.

cs.HC↗

arXiv2024

TopoCellGen: Generating Histopathology Cell Topology with a Diffusion Model

Accurately modeling multi-class cell topology is crucial in digital pathology, as it provides critical insights into tissue structure and pathology. The synthetic generation of cell topology enables realistic simulations of complex tissue environments, enhances downstream tasks by augmenting training data, aligns more closely with pathologists' domain knowledge, and offers new opportunities for controlling and generalizing the tumor microenvironment. In this paper, we propose a novel approach that integrates topological constraints into a diffusion model to improve the generation of realistic, contextually accurate cell topologies. Our method refines the simulation of cell distributions and interactions, increasing the precision and interpretability of results in downstream tasks such as cell detection and classification. To assess the topological fidelity of generated layouts, we introduce a new metric, Topological Frechet Distance (TopoFD), which overcomes the limitations of traditional metrics like FID in evaluating topological structure. Experimental results demonstrate the effectiveness of our approach in generating multi-class cell layouts that capture intricate topological relationships. Code is available at https://github.com/Melon-Xu/TopoCellGen.

eess.IV↗

arXiv2024

Touchstone Benchmark: Are We on the Right Way for Evaluating AI Algorithms for Medical Segmentation?

How can we test AI performance? This question seems trivial, but it isn't. Standard benchmarks often have problems such as in-distribution and small-size test sets, oversimplified metrics, unfair comparisons, and short-term outcome pressure. As a consequence, good performance on standard benchmarks does not guarantee success in real-world scenarios. To address these problems, we present Touchstone, a large-scale collaborative segmentation benchmark of 9 types of abdominal organs. This benchmark is based on 5,195 training CT scans from 76 hospitals around the world and 5,903 testing CT scans from 11 additional hospitals. This diverse test set enhances the statistical significance of benchmark results and rigorously evaluates AI algorithms across various out-of-distribution scenarios. We invited 14 inventors of 19 AI algorithms to train their algorithms, while our team, as a third party, independently evaluated these algorithms on three test sets. In addition, we also evaluated pre-existing AI frameworks--which, differing from algorithms, are more flexible and can support different algorithms--including MONAI from NVIDIA, nnU-Net from DKFZ, and numerous other open-source frameworks. We are committed to expanding this benchmark to encourage more innovation of AI algorithms for the medical domain.

cs.CV↗

arXiv2024

Twins in Diversity: Understanding circumstellar disk evolution in the twin clusters of W5 complex

Young star-forming regions in massive environments are ideal test beds to study the influence of surroundings on the evolution of disks around low-mass stars. We explore two distant young clusters, IC 1848-East and West located in the massive W5 complex. These clusters are unique due to their similar (distance, age, and extinction) yet distinct (stellar density and FUV radiation fields) physical properties. We use deep multi-band photometry in optical, near-IR, and mid-IR wavelengths complete down to the substellar limit in at least five bands. We trace the spectral energy distribution of the sources to identify the young pre-main sequence members in the region and derive their physical parameters. The disk fraction for the East and West clusters down to 0.1 M$_\odot$ was found to be $\sim$27$\pm$2% (N$_{disk}$=184, N$_{diskless}$=492) and $\sim$17$\pm$1% (N$_{disk}$=173, N$_{diskless}$=814), respectively. While no spatial variation in the disk fraction is observed, these values are lower than those in other nearby young clusters. Investigating the cause of this decrease, we find a correlation with the intense feedback from massive stars throughout the cluster area. We also identified the disk sources undergoing accretion and observed the mass accretion rates to exhibit a positive linear relationship with the stellar host mass and an inverse relationship with stellar age. Our findings suggest that the environment significantly influences the dissipation of disks in both clusters. These distant clusters, characterized by their unique attributes, can serve as templates for future studies in outer galaxy regions, offering insights into the influence of feedback mechanisms on star and planetary formation.

astro-ph.SR↗

arXiv2024

TopoDiffusionNet: A Topology-aware Diffusion Model

Diffusion models excel at creating visually impressive images but often struggle to generate images with a specified topology. The Betti number, which represents the number of structures in an image, is a fundamental measure in topology. Yet, diffusion models fail to satisfy even this basic constraint. This limitation restricts their utility in applications requiring exact control, like robotics and environmental modeling. To address this, we propose TopoDiffusionNet (TDN), a novel approach that enforces diffusion models to maintain the desired topology. We leverage tools from topological data analysis, particularly persistent homology, to extract the topological structures within an image. We then design a topology-based objective function to guide the denoising process, preserving intended structures while suppressing noisy ones. Our experiments across four datasets demonstrate significant improvements in topological accuracy. TDN is the first to integrate topology with diffusion models, opening new avenues of research in this area. Code available at https://github.com/Saumya-Gupta-26/TopoDiffusionNet

cs.CV↗

arXiv2024

Backdooring Vision-Language Models with Out-Of-Distribution Data

The emergence of Vision-Language Models (VLMs) represents a significant advancement in integrating computer vision with Large Language Models (LLMs) to generate detailed text descriptions from visual inputs. Despite their growing importance, the security of VLMs, particularly against backdoor attacks, is under explored. Moreover, prior works often assume attackers have access to the original training data, which is often unrealistic. In this paper, we address a more practical and challenging scenario where attackers must rely solely on Out-Of-Distribution (OOD) data. We introduce VLOOD (Backdooring Vision-Language Models with Out-of-Distribution Data), a novel approach with two key contributions: (1) demonstrating backdoor attacks on VLMs in complex image-to-text tasks while minimizing degradation of the original semantics under poisoned inputs, and (2) proposing innovative techniques for backdoor injection without requiring any access to the original training data. Our evaluation on image captioning and visual question answering (VQA) tasks confirms the effectiveness of VLOOD, revealing a critical security vulnerability in VLMs and laying the foundation for future research on securing multimodal models against sophisticated threats.

cs.CV↗

arXiv2024

A Comprehensive Convolutional Neural Network Architecture Design using Magnetic Skyrmion and Domain Wall

Spintronic-based neuromorphic hardware offers high-density and rapid data processing at nanoscale lengths by leveraging magnetic configurations like skyrmion and domain walls. Here, we present the maximal hardware implementation of a convolutional neural network (CNN) based on a compact multi-bit skyrmion-based synapse and a hybrid CMOS domain wall-based circuit for activation and max-pooling functionalities. We demonstrate the micromagnetic design and operation of a circular bilayer skyrmion system mimicking a scalable artificial synapse, demonstrated up to 6-bit (64 states) with an ultra-low energy consumption of 0.87 fJ per state update. We further show that the synaptic weight modulation is achieved by the perpendicular current interaction with the labyrinth-maze like uniaxial anisotropy profile, inducing skyrmionic gyration, thereby enabling long-term potentiation (LTP) and long-term depression (LTD) operations. Furthermore, we present a simultaneous rectified linear (ReLU) activation and max pooling circuitry featuring a SOT-based domain wall ReLU with a power consumption of 4.73 $\mu$W. The ReLU function, stabilized by a parabolic uniaxial anisotropy profile, encodes domain wall positions into continuous resistance states coupled with the HSPICE circuit simulator. Our integrated skyrmion and domain wall-based spintronic hardware achieves 98.07% accuracy in convolutional neural network (CNN) based pattern recognition task, consuming 110 mW per image.

cond-mat.mes-hall↗

arXiv2024

Ultrahigh Frequency and Multi-channel Output in Skyrmion Based Nano-oscillator

Spintronic nano-oscillators can generate tunable microwave signals that find a wide range of applications in the field of telecommunication to modern neuromorphic computing systems. Among other spintronic devices, a magnetic skyrmion is a promising candidate for the next generation of low-power devices due to its small size and topological stability. In this work, we propose a multi-channel oscillator design based on the synthetic anti-ferromagnetic (SAF) skyrmion pair. The mitigation of the skyrmion Hall effect in SAF and the associated decimation of the Magnus force endows the proposed oscillator with an ultra-high frequency of 41GHz and a multi-channel frequency output driven by the same current. The ultrahigh operational frequency represents an $\sim$342 times improvement compared to the monolayer single skyrmion oscillator featuring a constant uniaxial anisotropy profile. Using micromagnetic simulations, we demonstrate the effectiveness of our proposed multi-channel oscillator design by introducing multi-channel nanotracks along with multiple skyrmions for enhanced frequency operation. The ultrahigh operational frequency and multi-channel output are attributed to three key factors: The oscillator design accounting for a finite spin-flip length of the spacer (such as Ru) material, tangential velocity proportionality on input spin current along with weak dependence on the radius of rotation of the skyrmion-pair, skyrmion interlocking in the channel enabled by the multi-channel high Ku rings and skyrmion-skyrmion repulsion, therefore resulting ultrahigh frequency and multi-channel outputs.

cond-mat.mes-hall↗