arXiv subjects

Ayan Banerjee

Publications and source records attributed to Ayan Banerjee.

At least 37 records · Page 2Linked to original sources

arXiv2026

Hardware Acceleration for Neural Networks: A Comprehensive Survey

Neural networks have become dominant computational workloads across cloud and edge platforms, but their rapid growth in model size and deployment diversity has exposed hardware bottlenecks increasingly dominated by memory movement, communication, and irregular operators rather than peak arithmetic throughput. This survey reviews the current technology landscape for hardware acceleration of deep learning, spanning GPUs and tensor-core architectures, domain-specific accelerators (TPUs, NPUs), FPGA-based designs, ASIC inference engines, and emerging LLM-serving accelerators such as LPUs, alongside in-/near-memory computing and neuromorphic/analog approaches. We organize the survey using a unified taxonomy across (i) workloads (CNNs, RNNs, GNNs, Transformers/LLMs), (ii) execution settings (training vs.\ inference; datacenter vs.\ edge), and (iii) optimization levers (reduced precision, sparsity and pruning, operator fusion, compilation and scheduling, memory-system/interconnect design). We synthesize key architectural ideas such as systolic arrays, vector and SIMD engines, specialized attention and softmax kernels, quantization-aware datapaths, and high-bandwidth memory, and discuss how software stacks and compilers bridge model semantics to hardware. Finally, we highlight open challenges -- including efficient long-context LLM inference (KV-cache management), robust support for dynamic and sparse workloads, energy- and security-aware deployment, and fair benchmarking -- pointing to promising directions for the next generation of neural acceleration.

eess.SY↗

arXiv2026

Personalized Model-Based Design of Human Centric AI enabled CPS for Long term usage

Human centric critical systems are increasingly involving artificial intelligence to enable knowledge extraction from sensor collected data. Examples include medical monitoring and control systems, gesture based human computer interaction systems, and autonomous cars. Such systems are intended to operate for a long term potentially for a lifetime in many scenarios such as closed loop blood glucose control for Type 1 diabetics, self-driving cars, and monitoting systems for stroke diagnosis, and rehabilitation. Long term operation of such AI enabled human centric applications can expose them to corner cases for which their operation is may be uncertain. This can be due to many reasons such as inherent flaws in the design, limited resources for testing, inherent computational limitations of the testing methodology, or unknown use cases resulting from human interaction with the system. Such untested corner cases or cases for which the system performance is uncertain can lead to violations in the safety, sustainability, and security requirements of the system. In this paper, we analyze the existing techniques for safety, sustainability, and security analysis of an AI enabled human centric control system and discuss their limitations for testing the system for long term use in practice. We then propose personalized model based solutions for potentially eliminating such limitations.

cs.AI↗

arXiv2026

Detection of Deployment Operational Deviations for Safety and Security of AI-Enabled Human-Centric Cyber Physical Systems

In recent years, Human-centric cyber-physical systems have increasingly involved artificial intelligence to enable knowledge extraction from sensor-collected data. Examples include medical monitoring and control systems, as well as autonomous cars. Such systems are intended to operate according to the protocols and guidelines for regular system operations. However, in many scenarios, such as closed-loop blood glucose control for Type 1 diabetics, self-driving cars, and monitoring systems for stroke diagnosis. The operations of such AI-enabled human-centric applications can expose them to cases for which their operational mode may be uncertain, for instance, resulting from the interactions with a human with the system. Such cases, in which the system is in uncertain conditions, can violate the system's safety and security requirements. This paper will discuss operational deviations that can lead these systems to operate in unknown conditions. We will then create a framework to evaluate different strategies for ensuring the safety and security of AI-enabled human-centric cyber-physical systems in operation deployment. Then, as an example, we show a personalized image-based novel technique for detecting the non-announcement of meals in closed-loop blood glucose control for Type 1 diabetics.

cs.CV↗

arXiv2026

XAI-MeD: Explainable Knowledge Guided Neuro-Symbolic Framework for Domain Generalization and Rare Class Detection in Medical Imaging

Explainability domain generalization and rare class reliability are critical challenges in medical AI where deep models often fail under real world distribution shifts and exhibit bias against infrequent clinical conditions This paper introduces XAIMeD an explainable medical AI framework that integrates clinically accurate expert knowledge into deep learning through a unified neuro symbolic architecture XAIMeD is designed to improve robustness under distribution shift enhance rare class sensitivity and deliver transparent clinically aligned interpretations The framework encodes clinical expertise as logical connectives over atomic medical propositions transforming them into machine checkable class specific rules Their diagnostic utility is quantified through weighted feature satisfaction scores enabling a symbolic reasoning branch that complements neural predictions A confidence weighted fusion integrates symbolic and deep outputs while a Hunt inspired adaptive routing mechanism guided by Entropy Imbalance Gain EIG and Rare Class Gini mitigates class imbalance high intra class variability and uncertainty We evaluate XAIMeD across diverse modalities on four challenging tasks i Seizure Onset Zone SOZ localization from rs fMRI ii Diabetic Retinopathy grading across 6 multicenter datasets demonstrate substantial performance improvements including 6 percent gains in cross domain generalization and a 10 percent improved rare class F1 score far outperforming state of the art deep learning baselines Ablation studies confirm that the clinically grounded symbolic components act as effective regularizers ensuring robustness to distribution shifts XAIMeD thus provides a principled clinically faithful and interpretable approach to multimodal medical AI.

cs.AI↗

arXiv2025

Enabling Physical AI at the Edge: Hardware-Accelerated Recovery of System Dynamics

Physical AI at the edge -- enabling autonomous systems to understand and predict real-world dynamics in real time -- requires hardware-efficient learning and inference. Model recovery (MR), which identifies governing equations from sensor data, is a key primitive for safe and explainable monitoring in mission-critical autonomous systems operating under strict latency, compute, and power constraints. However, state-of-the-art MR methods (e.g., EMILY and PINN+SR) rely on Neural ODE formulations that require iterative solvers and are difficult to accelerate efficiently on edge hardware. We present \textbf{MERINDA} (Model Recovery in Reconfigurable Dynamic Architecture), an FPGA-accelerated MR framework designed to make physical AI practical on resource-constrained devices. MERINDA replaces expensive Neural ODE components with a hardware-friendly formulation that combines (i) GRU-based discretized dynamics, (ii) dense inverse-ODE layers, (iii) sparsity-driven dropout, and (iv) lightweight ODE solvers. The resulting computation is structured for streaming parallelism, enabling critical kernels to be fully parallelized on the FPGA. Across four benchmark nonlinear dynamical systems, MERINDA delivers substantial gains over GPU implementations: \textbf{114$\times$ lower energy} (434~J vs.\ 49{,}375~J), \textbf{28$\times$ smaller memory footprint} (214~MB vs.\ 6{,}118~MB), and \textbf{1.68$\times$ faster training}, while matching state-of-the-art model-recovery accuracy. These results demonstrate that MERINDA can bring accurate, explainable MR to the edge for real-time monitoring of autonomous systems.

cs.LG↗

arXiv2025

NEURO-GUARD: Neuro-Symbolic Generalization and Unbiased Adaptive Routing for Diagnostics -- Explainable Medical AI

Accurate yet interpretable image-based diagnosis remains a central challenge in medical AI, particularly in settings characterized by limited data, subtle visual cues, and high-stakes clinical decision-making. Most existing vision models rely on purely data-driven learning and produce black-box predictions with limited interpretability and poor cross-domain generalization, hindering their real-world clinical adoption. We present NEURO-GUARD, a novel knowledge-guided vision framework that integrates Vision Transformers (ViTs) with language-driven reasoning to improve performance, transparency, and domain robustness. NEURO-GUARD employs a retrieval-augmented generation (RAG) mechanism for self-verification, in which a large language model (LLM) iteratively generates, evaluates, and refines feature-extraction code for medical images. By grounding this process in clinical guidelines and expert knowledge, the framework progressively enhances feature detection and classification beyond purely data-driven baselines. Extensive experiments on diabetic retinopathy classification across four benchmark datasets APTOS, EyePACS, Messidor-1, and Messidor-2 demonstrate that NEURO-GUARD improves accuracy by 6.2% over a ViT-only baseline (84.69% vs. 78.4%) and achieves a 5% gain in domain generalization. Additional evaluations on MRI-based seizure detection further confirm its cross-domain robustness, consistently outperforming existing methods. Overall, NEURO-GUARD bridges symbolic medical reasoning with subsymbolic visual learning, enabling interpretable, knowledge-aware, and generalizable medical image diagnosis while achieving state-of-the-art performance across multiple datasets.

cs.AI↗

arXiv2025

Accelerated Digital Twin Learning for Edge AI: A Comparison of FPGA and Mobile GPU

Digital twins (DTs) can enable precision healthcare by continually learning a mathematical representation of patient-specific dynamics. However, mission critical healthcare applications require fast, resource-efficient DT learning, which is often infeasible with existing model recovery (MR) techniques due to their reliance on iterative solvers and high compute/memory demands. In this paper, we present a general DT learning framework that is amenable to acceleration on reconfigurable hardware such as FPGAs, enabling substantial speedup and energy efficiency. We compare our FPGA-based implementation with a multi-processing implementation in mobile GPU, which is a popular choice for AI in edge devices. Further, we compare both edge AI implementations with cloud GPU baseline. Specifically, our FPGA implementation achieves an 8.8x improvement in \text{performance-per-watt} for the MR task, a 28.5x reduction in DRAM footprint, and a 1.67x runtime speedup compared to cloud GPU baselines. On the other hand, mobile GPU achieves 2x better performance per watts but has 2x increase in runtime and 10x more DRAM footprint than FPGA. We show the usage of this technique in DT guided synthetic data generation for Type 1 Diabetes and proactive coronary artery disease detection.

cs.DC↗

arXiv2025

Fast Online Digital Twinning on FPGA for Mission Critical Applications

Digital twinning enables real-time simulation and predictive modeling by maintaining a continuously updated virtual representation of a physical system. In mission-critical applications, such as mid-air collision avoidance, these models must operate online with extremely low latency to ensure safety. However, executing complex Model Recovery (MR) pipelines on edge devices is limited by computational and memory bandwidth constraints. This paper introduces a fast, FPGA-accelerated digital twinning framework that offloads key neural components, including gated recurrent units (GRU) and dense layers, to reconfigurable hardware for efficient parallel execution. Our system achieves real-time responsiveness, operating five times faster than typical human reaction time, and demonstrates the practical viability of deploying digital twins on edge platforms for time-sensitive, safety-critical environments.

cs.DC↗

arXiv2025

Photophoretic Trapping: Fundamentals, Advances and Future Directions

Photophoretic forces, several orders of magnitude stronger than radiation pressure, enable particle trapping at remarkably low optical intensities and have opened pathways to applications in aerosol science, free-space 3D volumetric displays, and even deployment of lightweight payloads in space. In this review, we provide a comprehensive explanation of the underlying physics of photophoretic forces and how they facilitate stable three-dimensional manipulation of absorbing particles. We examine the experimental configurations that enable robust trapping, and we detail the physical parameters that govern the magnitude and behavior of photophoretic forces in these geometries. The rich dynamical phenomena exhibited by photophoretically trapped particles are discussed alongside current and emerging applications and possible future research directions. This review thus attempts to systematically unify the theoretical, experimental, and application-oriented aspects of photophoretic trapping, with the aim of advancing and strengthening research in this rapidly developing field.

physics.optics↗

arXiv2025

MedXAI: A Retrieval-Augmented and Self-Verifying Framework for Knowledge-Guided Medical Image Analysis

Accurate and interpretable image-based diagnosis remains a fundamental challenge in medical AI, particularly under domain shifts and rare-class conditions. Deep learning models often struggle with real-world distribution changes, exhibit bias against infrequent pathologies, and lack the transparency required for deployment in safety-critical clinical environments. We introduce MedXAI (An Explainable Framework for Medical Imaging Classification), a unified expert knowledge based framework that integrates deep vision models with clinician-derived expert knowledge to improve generalization, reduce rare-class bias, and provide human-understandable explanations by localizing the relevant diagnostic features rather than relying on technical post-hoc methods (e.g., Saliency Maps, LIME). We evaluate MedXAI across heterogeneous modalities on two challenging tasks: (i) Seizure Onset Zone localization from resting-state fMRI, and (ii) Diabetic Retinopathy grading. Ex periments on ten multicenter datasets show consistent gains, including a 3% improvement in cross-domain generalization and a 10% improvmnet in F1 score of rare class, substantially outperforming strong deep learning baselines. Ablations confirm that the symbolic components act as effective clinical priors and regularizers, improving robustness under distribution shift. MedXAI delivers clinically aligned explanations while achieving superior in-domain and cross-domain performance, particularly for rare diseases in multimodal medical AI.

cs.LG↗

arXiv2025

Hardware Software Optimizations for Fast Model Recovery on Reconfigurable Architectures

Model Recovery (MR) is a core primitive for physical AI and real-time digital twins, but GPUs often execute MR inefficiently due to iterative dependencies, kernel-launch overheads, underutilized memory bandwidth, and high data-movement latency. We present MERINDA, an FPGA-accelerated MR framework that restructures computation as a streaming dataflow pipeline. MERINDA exploits on-chip locality through BRAM tiling, fixed-point kernels, and the concurrent use of LUT fabric and carry-chain adders to expose fine-grained spatial parallelism while minimizing off-chip traffic. This hardware-aware formulation removes synchronization bottlenecks and sustains high throughput across the iterative updates in MR. On representative MR workloads, MERINDA delivers up to 6.3x fewer cycles than an FPGA-based LTC baseline, enabling real-time performance for time-critical physical systems.

cs.AR↗

arXiv2025

Adaptive Optics-Enhanced Michelson Interferometer for Spectroscopy of Narrow-Band Light Sources

Adaptive optics enables the deployment of interferometer-based spectroscopy without the need for moving parts necessary for scanning the interferometer arms. Here, we employ a Michelson Interferometer in conjunction with a Spatial Light Modulator (SLM) for determining the spectral profile of a narrow-band light source. Interestingly, we observe that the fringes across the interferometer output beam are inherently shifted in wavelength even when a constant phase profile is provided to the SLM. We calibrate the spectral shifts as a function of fringe spatial location by measuring the incident light spectrum at various points across the fringe pattern, and observe that the spectral peak traces out a `teardrop' shape, whose width is dependent on the spectral bandwidth of the source, the relative tilt and path difference between the two arms of the interferometer, and the divergence of the beam. Next, we demonstrate that this inherent spectral variation of the fringes can be used to perform fast single-snapshot spectroscopy of narrow-band light sources, while a time-varied phase profile provided to the SLM leads to multi-step spectroscopy with lower noise, higher resolution, and better contrast. Our findings establish that the Michelson Interferometer can be used to perform spectroscopy of any source within a certain spectral range from simple images of the fringe pattern, so as to facilitate exciting applications towards hyperspectral imaging.

physics.optics↗

arXiv2025

Model Recovery at the Edge under Resource Constraints for Physical AI

Model Recovery (MR) enables safe, explainable decision making in mission-critical autonomous systems (MCAS) by learning governing dynamical equations, but its deployment on edge devices is hindered by the iterative nature of neural ordinary differential equations (NODEs), which are inefficient on FPGAs. Memory and energy consumption are the main concerns when applying MR on edge devices for real-time operation. We propose MERINDA, a novel FPGA-accelerated MR framework that replaces iterative solvers with a parallelizable neural architecture equivalent to NODEs. MERINDA achieves nearly 11x lower DRAM usage and 2.2x faster runtime compared to mobile GPUs. Experiments reveal an inverse relationship between memory and energy at fixed accuracy, highlighting MERINDA's suitability for resource-constrained, real-time MCAS.

cs.AI↗

arXiv2025

CraftSVG: Multi-Object Text-to-SVG Synthesis via Layout Guided Diffusion

Generating VectorArt from text prompts is a challenging vision task, requiring diverse yet realistic depictions of the seen as well as unseen entities. However, existing research has been mostly limited to the generation of single objects, rather than comprehensive scenes comprising multiple elements. In response, this work introduces SVGCraft, a novel end-to-end framework for the creation of vector graphics depicting entire scenes from textual descriptions. Utilizing a pre-trained LLM for layout generation from text prompts, this framework introduces a technique for producing masked latents in specified bounding boxes for accurate object placement. It introduces a fusion mechanism for integrating attention maps and employs a diffusion U-Net for coherent composition, speeding up the drawing process. The resulting SVG is optimized using a pre-trained encoder and LPIPS loss with opacity modulation to maximize similarity. Additionally, this work explores the potential of primitive shapes in facilitating canvas completion in constrained environments. Through both qualitative and quantitative assessments, SVGCraft is demonstrated to surpass prior works in abstraction, recognizability, and detail, as evidenced by its performance metrics (CLIP-T: 0.4563, Cosine Similarity: 0.6342, Confusion: 0.66, Aesthetic: 6.7832). The code will be available at https://github.com/ayanban011/SVGCraft.

cs.CV↗

arXiv2025

Surface Plasmon Mediated Giant Goos-Hanchen and Imbert-Fedorov Shifts on a Corrugated Metal Surface

Enhanced beam shifts mediated by surface plasmon resonance (SPR) at metal-dielectric interfaces have been widely investigated. However, research on the associated Imbert-Fedorov or spin Hall shifts, driven by the spin-orbit interaction of structured light in structured interfaces, has been comparatively scarce and limited. We explore the reflection characteristics of generic polarized, non-paraxial light beams from a corrugated silver (Ag) interface, since surface corrugation can naturally couple the incident radiation modes to the surface excitations. In the vicinity of SPR, we report a significant enhancement in the beam shifts, attributed to the rapid variation of the specular reflection coefficient near its minima, resulting in amplified weak values. By carefully selecting the incident and projected polarization states of the beam, we achieve a pronounced spatial spin Hall effect. We also investigate vortex-induced beam shifts within this resonant regime, revealing distinctive signatures of the angular momentum of the beam. Furthermore, a comprehensive analysis is also presented for the conical diffraction geometry, wherein polarization conversions between p and s states are fully incorporated. Our work establishes the interplay of the spin-orbit interaction of light and the weak measurement approach as an important methodology in amplifying SPR effects, which may have important connotations in applications involving light at nanoscales.

physics.optics↗

arXiv2025

Dark-Energy Anisotropic Compact Configurations in 4D Einstein-Gauss-Bonnet Gravity: From Structure to Observational Viability

We address the equilibrium configurations and stability properties of anisotropic compact stars whose interior is described by a modified Chaplygin gas (MCG) equation of state in the framework of the regularized four-dimensional Einstein-Gauss-Bonnet (4DEGB) theory. Applying a quasi-local prescription for the pressure anisotropy, we derive the modified Tolman-Oppenheimer-Volkoff (TOV) equations and integrate them numerically over a large parameter space in the Gauss-Bonnet coupling $α$ and the degree of anisotropy $β$. We provide mass-radius sequences, mass-compactness, energy density, and pressure profiles, and perform a full stability analysis based on the turning-point criterion, the radial adiabatic index $γ_r$, and the radial and transverse sound speeds $v_r^2$ and $v_t^2$. Our results show that positive $α$ and positive anisotropy $(β> 0)$ systematically increase the maximum mass and radius, enabling then configurations that exceed $2\,M_\odot$ while still obeying causality and the modified Buchdahl bound in 4DEGB gravity. A comparison with the latest astrophysical constraints (NICER, GW170817, GW190814, and massive-pulsar measurements) identifies regions of the $(α,β)$ parameter space that are observationally allowable. In conclusion, anisotropic dark-energy stars in 4DEGB gravity provide viable, observationally testable ultra-compact alternatives to normal neutron stars and black holes, and also potentially open rich avenues for further multi-messenger searches for higher-curvature effects.

gr-qc↗

arXiv2025

Spin-Split Dispersion of Leaky Surface plasmons in Inversion- Symmetric System

Spin-dependent dispersion and Rashba effect are manifestations of universal spin orbit interaction associated with the breaking of the spatial inversion symmetry in condensed matter and in optical systems. In sharp contrast to this, we report a spin-split dispersion effect of leaky surface plasmons in an inversion-symmetric one dimensional plasmonic grating system. In our system, the signature of spin-momentum locking and the resulting spin-polarization dependent splitting of dispersion of the surface plasmons are observed through the leakage radiation detected in a Fourier (momentum) domain optical arrangement. The setup enables single-shot recording of the full polarization-resolved dispersion (frequency vs transverse momentum (k)) of the leaky surface plasmons. Momentum domain polarization analysis identified a transverse momentum (k) dependent linear birefringence-linear dichroism effect (referred to as the geometric LB-LD effect) responsible for the observed spin-split dispersion. This unconventional SOI effect is reminiscent of the recently reported LB-LD effect resulting in giant chirality in centrosymmetric crystal, albeit with geometric origin. It is demonstrated that the interplay of the geometrical polarization transformation in focused polarized light and subsequent interaction of the structured field polarization with the plasmonic grating leads to the evolution of strong geometrical phase gradient or spin(circular polarization)-dependent transverse momentum of light resulting in spin-split dispersion. Our study offers a new paradigm of spin-based dispersion engineering and spin-enabled nano-optical functionalities in simple symmetric metasurfaces using geometric LB-LD effect.

physics.optics↗

arXiv2025

Dark Energy Stars in Rastall-Rainbow Gravity: Structure, Stability and Observational Constraints

In this work, we investigate static configurations of dark energy stars within the framework of Rastall-Rainbow (R-R) gravity, which combines an energy-dependent deformation of spacetime with a nonminimal coupling between matter and geometry. We begin by deriving the modified field equations corresponding to R-R gravity and subsequently reformulate the stellar structure equations to describe hydrostatic equilibrium. The generalized Tolman-Oppenheimer-Volkoff (TOV) equations are then solved numerically by adopting the modified Chaplygin equation of state to model the interior matter distribution. The R-R parameters, along with fluid constants, are shown to influence the maximum mass, radii, and stiffness of the star sequences compared to the baseline set by general relativity. We apply observational benchmarks from high-mass pulsars and binary-merger events (e.g., GW170817 and GW190814) to appraise viability within the explored parameter space. The results collectively suggest that stable, causal configurations arise from physically meaningful parameter selections, with deviations from general relativity leading to systematic changes in structural characteristics while adhering to theoretical limits. These findings illustrate that Rastall-Rainbow gravity can support stable, observationally consistent dark energy stars, providing verifiable signatures in strong gravitational fields.

gr-qc↗