arXiv subjects

Soumya Chatterjee

Publications and source records attributed to Soumya Chatterjee.

At least 19 recordsLinked to original sources

arXiv2026

HierarchicalDAEW: Domain-Aware Edge-Weighted Graph Convolution with Evidential Uncertainty for Multi-Section Spatial Gene Expression Prediction from H&E Histology

Spatial transcriptomics assays remain costly and technically demanding, restricting transcriptome-wide profiling to specialist settings and preventing routine clinical deployment. Predicting spatially resolved gene expression from H&E histology could close this gap, yet current methods largely ignore the underlying tissue architecture and rarely quantify how their predictions can be trusted. We introduce HierarchicalDAEW, a dual-graph architecture that addresses both gaps. On the spot graph, a Domain-Aware Edge-Weighted convolutional operator learns separate projections for inter-domain, intra-domain, and boundary edges derived from Leiden clustering, allowing the model to treat tissue heterogeneity as an explicit structural signal rather than an implicit one. A second gene-level graph then fuses protein-protein interaction priors from STRING-DB with tissue-specific co-expression through learned attention gating, propagating predictions from a landmark gene set to a broader gene panel. Reliability is handled through evidential uncertainty estimation, which produces far better calibrated confidence intervals than Monte Carlo dropout under identical conditions. Across six human Visium sections spanning breast, colorectal, prostate, and cerebellar tissue, and against thirteen published baselines, HierarchicalDAEW achieves the strongest correlation with ground-truth expression, with gains that hold up under multi-seed reproducibility checks and negative controls that rule out positional shortcuts. Ablations further confirm that both the domain-aware edge typing and the hierarchical depth are necessary to this improvement, and calibrated uncertainty estimates identify low-confidence predictions for pathologist review before clinical action.

cs.LG↗

arXiv2026

Counterfactual Explainability Framework With CycleGAN And Counterfactual-Classifier Alignnment Score for Retinal Disease Classification

Automated detection of vision impairing retina-based ocular conditions from fundus images is important for early screening, timely referral and reducing dependency on specialist-only assessment, for which neural network-based deep learning (DL) models have been widely utilized. However, explainability of the DL frameworks remains a major bottleneck for clinical adoption, particularly when model decisions are not linked to retinal regions that are clinically meaningful. To address this issue, this study presents CounterFundus, a novel CycleGAN-driven counterfactual explainability framework, integrating EfficientNet-B5-based retinal disease detection with visually interpretable disease-to-normal fundus image translation. For each pathological image, the counterfactual yielded by the CycleGAN generator represents an estimated healthy counterpart and the resultant difference map is utilized to localize disease-associated retinal changes. Unlike conventional post-hoc saliency methods, CounterFundus provides counterfactual explanations through visually plausible disease-to-normal retinal translation. Thereafter, to quantify the spatial agreement between counterfactual difference maps and classifier saliency, the Counterfactual-Classifier Alignment Score (CCAS) is introduced, embedding Spearman correlation, binary IoU and pointing accuracy into a single assessment protocol. To this end, EigenCAM-aligned evaluation demonstrates that the generated counterfactual explanations remain spatially consistent with classifier-relevant retinal evidence across all CCAS dimensions. Along with that, ablation studies further confirm that CCAS-filtered counterfactual augmentation improves the downstream classification performance in fundus images, establishing CounterFundus as a clinically-grounded, explainable artificially intelligence (XAI) framework for retinal disease detection.

cs.LG↗

arXiv2025

PORTool: Importance-Aware Policy Optimization with Rewarded Tree for Multi-Tool-Integrated Reasoning

Multi-tool-integrated reasoning enables LLM-empowered tool-use agents to solve complex tasks by interleaving natural-language reasoning with calls to external tools. However, training such agents from outcome-only rewards suffers from credit-assignment ambiguity, obscuring which intermediate tool-use decisions drive success or failure. In this paper, we propose PORTool, an importance-aware policy-optimization algorithm that reinforces agents' tool-use competence from outcome-level supervision while assigning reward at the step level. Specifically, PORTool generates a rewarded rollout tree in which trajectories share prefixes before branching, enabling direct comparisons among alternative tool-use decisions within the same context. It then estimates each step's importance by a correctness-dominant signal, i.e., whether descendants of that step can ultimately produce a correct final answer, plus an auxiliary term indicating whether the step's tool calls satisfy formatting constraints and execute successfully. Using these step-wise importance estimates, PORTool updates the policy to generate efficient tool-call steps, guided by both local comparisons within each branching decision and the overall quality of entire trajectories. Experiments show that PORTool improves final-answer accuracy while reducing tool-call steps compared with state-of-the-art policy-optimization baselines, and ablation studies confirm the robustness of the proposed step-wise importance estimates.

cs.CL↗

arXiv2025

Zeptosecond to attosecond dynamics in atoms and possibility of generating a zeptosecond light source

In nuclear collisions, nuclear bremsstrahlung can cause nuclear Coulomb excitation via photon exchange in the projectile as well as the target nuclei. Such a process originating in nuclear timescales (zeptoseconds) can also influence the atomic phenomenon, which can be observed if it is delayed at least by a few attoseconds as atomic timescales $\ge$ an attosecond. We have found that this may happen due to a mechanism called the Eisenbud-Wigner-Smith (EWS)time delay process. We have estimated EWS time delays in atomic collisions utilizing the non-relativistic version of random phase approximation with exchange as well as Hartree-Fock methods. We present three representative collision systems through which one can experimentally observe the phenomena in attosecond timescales even though they originate from nuclear bremsstrahlung radiation occurring in zeptoseconds. Thus the present work represents an investigation of parallels between two neighboring areas of physics: atomic and nuclear physics. Furthermore the present work suggests the possibilities for atomic physics research near the Coulomb barrier energies, where the nuclear bremsstrahlung can be used as a zeptosecond x-ray source.

physics.atom-ph↗

arXiv2025

Towards full understanding the physics of heavy ion induced inner shell ionization

The complex physics of inner shell ionization of target atoms by heavy ion impact has remained only partially solved for decades. Recently, agreement between theory and experiment has been achieved by considering inner shell ionization of target atoms due to projectile electron capture in addition to direct Coulomb ionization including multiple ionization effects. A thorough investigation exhibits such a picture only if the atomic parameters of the target atoms are correct. In fact, the theoretical approach is found to be right, but the problem arises with the faulty atomic parameters. Furthermore, we show that fluorescence yields play a major role among the atomic parameters. We explore such a powerful method that enables us to measure the correct and accurate fluorescence yields for almost every element in the periodic table. As per our present knowledge, this in turn not only solves the said complex issue fully but also makes the PIXE analysis more reliable and accurate using both light and heavy ions.

physics.atom-ph↗

arXiv2025

Bremsstrahlung induced atomic processes

The observed spectra in the collisions of neon (Ne) projectiles of 1.8 and 2.1 MeV with an aluminum target (Al) have been successfully segregated from strong bremsstrahlung backgrounds and then analyzed by comparing the transition energies and rates with the theoretical predictions of the flexible atomic structure code and the general purpose relativistic atomic structure package. The spectra contain K{\alpha}, Kh{\alpha} , and K{\alpha}{\alpha} lines. The K{\alpha}{\alpha} emissions are due to two-electron one-photon transitions. Interestingly, the K{\alpha}{\alpha} lines in projectile ions are only seen with 1.8 MeV energy. In contrast, the K{\alpha}{\alpha} lines in the target ions are also well observed with 2.1 MeV energy. Surprisingly, the Al K x-ray line intensities are strongly suppressed, and the K{\alpha}{\alpha} line intensities are unexpectedly enhanced. The underlying physical process is found to be the photoionization caused by intense bremsstrahlung radiation. This photoionization process converts most of the singly ionized K shell states (\approx 82\% at 2.1 MeV) to doubly ionized K shell states. This phenomenon is silently present on many occasions. We take some of such events to validate this remarkable finding. This bremsstrahlung radiation induced secondary ionization process stands as an eye opening incidence to the plasma physics, astronomy and astrophysics communities; may revolutionize these fields of research.

physics.atom-ph↗

arXiv2024

Theoretical studies on multiple ionisation and electron capture processes in heavy ion induced M-shell ionisation

Multiple ionisation and electron capture are found to be vital mechanisms for K and L x-ray emissions along with the direct coulomb ionisation in heavy ion-atom collisions. Naturally, these two mechanisms may also be significant for M x-ray emissions also. However, these mechanisms associating with the M-shell are highly complex and not yet studied convincingly. We, in this work, have discussed about theoretical techniques to study these mechanisms in more general way, which in turn, almost resolved the wide gap between the theories and experiments regarding M-shell prevalent in literature up to a good extent. We have justified this technique with a few representative cases; for example, the silicon and sulphur projectile ions colliding with the gold and bismuth targets at beam energies ranging from 5 to 10 MeV. Here, we see that the capture contribution is much higher than that coming from the multiple ionisation effect. The combined effects of direct coulomb ionisation associated with simultaneous multiple ionisation and electron capture give a better agreement with the measurements up to a good extent.

physics.atom-ph↗

arXiv2024

Study of ion induced Inner Shell Ionization cross section through electron capture mechanism

Electron Capture (EC) cross-section from K, L and M shells of the target atoms to the vacant K, L and M shells of the projectile ions have been calculated by deriving the accurate momentum transfer to the captured electrons for different charge states. Several correction factors like polarization correction, relativistic effects (R) of the target wave function, Coulomb-deflection factor (C) due to the effect of the repulsion between the projectile and the target nucleus, correction for projectile energy loss have been introduced. The mean charge state of the projectiles inside the target material has been estimated using suitable empirical models and the fractional charge state distribution has been calculated considering Lorentz distribution. Fractional distribution of charge state of the projectile ions is used to obtain the charge state contributions of the electron capture cross-sections. The effect of Simultaneous Multiple Ionization (SMI) has been considered in the theory of Direct Coulomb Ionization (DCI). The theoretically obtained total cross-sections have been compared with the experimental findings obtained from various literature. The computation scheme has been depicted through sample calculations of ionization cross-sections through electron capture mechanism.

physics.atom-ph↗

arXiv2023

Resources and Evaluations for Multi-Distribution Dense Information Retrieval

We introduce and define the novel problem of multi-distribution information retrieval (IR) where given a query, systems need to retrieve passages from within multiple collections, each drawn from a different distribution. Some of these collections and distributions might not be available at training time. To evaluate methods for multi-distribution retrieval, we design three benchmarks for this task from existing single-distribution datasets, namely, a dataset based on question answering and two based on entity matching. We propose simple methods for this task which allocate the fixed retrieval budget (top-k passages) strategically across domains to prevent the known domains from consuming most of the budget. We show that our methods lead to an average of 3.8+ and up to 8.0 points improvements in Recall@100 across the datasets and that improvements are consistent when fine-tuning different base retrieval models. Our benchmarks are made publicly available.

cs.IR↗

arXiv2023

Study of bremsstrahlung radiation coming from a transmission type x-ray generator and its application on EDXRF technique

We have demonstrated the use of the bremsstrahlung radiation in energy dispersive x-ray fluorescence technique as a tool to perform non-destructive elemental analysis of solid samples employed in inter-disciplinary science research. The bremsstrahlung radiation can be taken from a small, portable, transmission type x-ray generator. As our knowledge goes, till date, this is the first attempt to generate bremsstrahlung spectrum theoretically from a portable transmission type x-ray generator and compared it with the actual observation. Theoretically generated bremsstrahlung spectra are also found to be in good agreement with the experimentally observed spectra obtained with various operating anode voltages of the x-ray generator. A computer program has been developed utilizing few atomic parameters to obtain the elemental concentrations in the sample by a single run using the whole bremsstrahlung. The knowledge of incoming x-ray flux, geometry of experimental arrangements is not required in this technique. To validate the technique, we have taken two samples: an Indian one rupee coin of the year 2000 and a NIST made brass sample, whose compositions are well known. We used these samples to expose to the bremsstrahlung radiation, produced by the operating anode voltages 20, 25, and 30 KV. The relative concentrations of different elements are determined, which is in good agreement with the earlier results.

physics.ins-det↗

arXiv2022

Matching options to tasks using Option-Indexed Hierarchical Reinforcement Learning

The options framework in Hierarchical Reinforcement Learning breaks down overall goals into a combination of options or simpler tasks and associated policies, allowing for abstraction in the action space. Ideally, these options can be reused across different higher-level goals; indeed, such reuse is necessary to realize the vision of a continual learning agent that can effectively leverage its prior experience. Previous approaches have only proposed limited forms of transfer of prelearned options to new task settings. We propose a novel option indexing approach to hierarchical learning (OI-HRL), where we learn an affinity function between options and the items present in the environment. This allows us to effectively reuse a large library of pretrained options, in zero-shot generalization at test time, by restricting goal-directed learning to only those options relevant to the task at hand. We develop a meta-training loop that learns the representations of options and environments over a series of HRL problems, by incorporating feedback about the relevance of retrieved options to the higher-level goal. We evaluate OI-HRL in two simulated settings - the CraftWorld and AI2THOR environments - and show that we achieve performance competitive with oracular baselines, and substantial gains over a baseline that has the entire option pool available for learning the hierarchical policy.

cs.LG↗

arXiv2022

Accurate Online Posterior Alignments for Principled Lexically-Constrained Decoding

Online alignment in machine translation refers to the task of aligning a target word to a source word when the target sequence has only been partially decoded. Good online alignments facilitate important applications such as lexically constrained translation where user-defined dictionaries are used to inject lexical constraints into the translation model. We propose a novel posterior alignment technique that is truly online in its execution and superior in terms of alignment error rates compared to existing methods. Our proposed inference technique jointly considers alignment and token probabilities in a principled manner and can be seamlessly integrated within existing constrained beam-search decoding algorithms. On five language pairs, including two distant language pairs, we achieve consistent drop in alignment error rates. When deployed on seven lexically constrained translation tasks, we achieve significant improvements in BLEU specifically around the constrained positions.

cs.CL↗

arXiv2021

Zeptosecond dynamics in atoms: fact or fiction?

Photon exchange due to nuclear bremsstrahlung during nuclear collisions can cause Coulomb excitation in the projectile and the target nuclei. The corresponding process originated in nuclear timescales can also be observed in atomic phenomenon experimentally if it delayed by at least with an attosecond or longer timescales. We have found that this happens due to a mechanism involving the Eisenbud-Wigner-Smith time delay process. We have estimated photoionization time delays in atomic collisions utilizing the nonrelativistic version of random phase approximation with exchange and Hartree-Fock methods. We present three representative processes in which we can observe the phenomena in attosecond timescales even though they originate from excitations in the zeptosecond timescales. Thus the work represents an investigation of parallels between two neighboring areas of physics. Furthermore the present work suggests new possibilities for atomic physics research near the Coulomb barrier energy, where the laser is replaced by nuclear bremsstrahlung.

physics.atom-ph↗

arXiv2021

Significance of high charge state of projectile ions inside the target and its role on electron capture leading to target ionization phenomenon

The K x-ray spectra of different targets (Cu, Zn, and Ge) induced by 3 to 5 MeV/u Si projectile ions have been measured to determine the K-shell ionization cross-section. A significant difference is observed between the measurements and theoretical estimates, with the latter being about 50% below the experimental results. This underestimation} is attributed to the charge-exchange from target K-shell to projectile K- and L-shells. Such observation can only be possible if the projectile ions attain up to H- and He-like charge states. Corresponding projectile charge state fractions have been evaluated from the Lorentzian charge state distribution, where mean charge state is taken from the Fermi gas model [Phys. Rev. Lett. 30, 358 (1973)] and width from the Novikov and Teplova approach [Phys. Lett. A378, 1286-1289 (2014)]. The sum of the direct ionization cross-section and K-K + K-L capture cross-sections gives a good agreement with the measured cross-sections. Furthermore, we have validated this methodology with available data for Si-ion on Ti target. Such results may be useful in many solid target based applications.

physics.atom-ph↗

arXiv2021

Understanding the L-shell ionization mechanism through osmium atoms bombarded by 4-6 MeV/u fluorine ions

The L-subshell ionization mechanism is studied in an ultra-thin osmium target bombarded by 4-6 MeV/u fluorine ions. Multiple ionization effects in the collisions are considered through the change of fluorescence and Coster-Kronig yields while determining L-subshell ionization cross sections from L-line x-ray production cross sections. The L-subshell ionization, as well as L-shell x-ray production cross sections so obtained, are compared with various theoretical approximations. The Coulomb direct ionization contributions is studied by (i) the relativistic semi-classical approximations (RSCA), (ii) the shellwise local plasma approximation (SLPA), and (iii) the ECUSAR theory, along with the inclusion of the vacancy sharing among the subshells by the coupled-states model (CSM) and the electron capture (EC) by a standard formalism. We find that the ECUSAR-CSM-EC describes the measured excitation function curves the best. However, the theoretical calculations are still about a factor of two smaller than the measured values. Such differences are resolved by re-evaluating the fluorescence and the Coster-Kronig yields. This work demonstrates that, in the present energy range, the heavy-ion induced inner-shell ionization of heavy atoms can be understood by combining the basic mechanisms of the direct Coulomb ionization, the electron capture, the multiple ionization, and the vacancy sharing among subshells, together with optimized atomic parameters.

physics.atom-ph↗

arXiv2021

Exploring a novel model for projectile charge state distribution inside a solid-target

For the first time, we report a theoretical methodology to predict charge state distribution of projectile ions inside a solid-target. The method utilizes either a simple Fermi gas model or an ab initio theoretical method and a certain parameterization of width for the Lorentzian charge state distributions. Results obtained from the two approaches are comparable, but the former has a certain edge over the latter. The projectile charge state distribution inside a solid-target plays a significant role in estimating electron capture cross-sections and then to describe the observed K-shell ionization dynamics. The electron capture process plays a certain role in L-shell ionization dynamics too, but in a test case of Si on Au target the subshell charge sharing contributes a more vital role than the electron capture. Thus, we have validated the present model as a reliable as well as useful for many solid-target based applications viz. tumour therapy, biophysics, accelerators, material science etc.

physics.atom-ph↗

arXiv2021

Evaluation of accurate uncertainty of measurement in L subshell ionization cross-section

To have a better understanding of a physical process, a comparison of ex-perimental data with theoretical values is mandatory. The comparison ismeaningful if the uncertainty in the experiment is accounted well. However,it is seldom seen, especially for a complex phenomenon. We take a test casethrough L subshell ionization of atoms by particle impact. Experimentally,x-ray production cross-sections are measured, but ionization cross-sectionsare calculated theoretically. Furthermore, the uncertainty of the x-ray pro-duction cross-section is mainly statistics and detector-efficiency driven. Butionization cross-section involves many other factors because of the relation-ship between the production and ionization cross section, having wide uncer-tainty spectrum. Consequently, determining the measurement uncertainty inL subshell ionization cross-section is always difficult. We have studied thisissue in the simplest way, where the rule of weighted propagation of rela-tive uncertainty is utilised. We notice that larger uncertainties are involvedin atomic parameters relevant to L1(2s1/2) subshell than those associatedwith the other two L2(2p1/2) and L3(2p3/2) subshells. Hence, comparisonbetween theory and experiment would give higher emphasis onL2andL3subshell ionization cross sections. We believe this work aware us that theappropriate uncertainty evaluation is extremely important for providing theright judgment on the data.

physics.atom-ph↗

arXiv2021

Model-agnostic Fits for Understanding Information Seeking Patterns in Humans

In decision making tasks under uncertainty, humans display characteristic biases in seeking, integrating, and acting upon information relevant to the task. Here, we reexamine data from previous carefully designed experiments, collected at scale, that measured and catalogued these biases in aggregate form. We design deep learning models that replicate these biases in aggregate, while also capturing individual variation in behavior. A key finding of our work is that paucity of data collected from each individual subject can be overcome by sampling large numbers of subjects from the population, while still capturing individual differences. In addition, we can predict human behavior with high accuracy without making any assumptions about task goals, reward structure, or individual biases, thus providing a model-agnostic fit to human behavior in the task. Such an approach can sidestep potential limitations in modeler-specified inductive biases, and has implications for computational modeling of human cognitive function in general, and of human-AI interfaces in particular.

cs.AI↗