Searcharxiv⌕ Search

arXiv subjects

Ayush Kumar

Publications and source records attributed to Ayush Kumar.

At least 19 recordsLinked to original sources
arXiv2026

Explaining Intrusion Alert Decisions of Deep Learning-based Network Intrusion Detection Systems for Security Analysts

In this paper, we present EXP-SEC, a novel framework which can explain the intrusion detection decisions of DL-based NIDS (which lead to security alerts) in a way that is aligned with the domain knowledge of analysts working in Security Operations Center (SOC). We highlight the following features of our framework: (1) a forensic module that isolates the suspect packets/flow which likely caused an alert (2) an explanation module which can handle much more complex feature dependencies in network traffic than existing methods (features can be divided into overlapping groups and some groups are more important than others), and (3) a multi-stage mapping module which translates the feature/group-based explanations generated by explanation module to domain-specific explanations suitable for processing by security analysts. We evaluate EXP-SEC with state-of-the-art DL-based NIDS and our evaluation results show that EXP-SEC outperforms xNIDS (existing best performing explanation framework) in terms of group-level and overlap-aware explanation utility metrics while performing similarly in terms of conventional feature-level metrics such as descriptive accuracy, sparsity and stability. Moreover, taking the case of a state-of-the-art DL-based NIDS, we demonstrate the security analyst-friendly explanation format generated by EXP-SEC.

cs.CR↗
arXiv2026

Prototype-Guided Latent Alignment for Data-Efficient Fine-Tuning of Molecular Foundation Models

Machine learning interatomic potentials (MLIPs) have transformed materials discovery by leveraging graph neural networks (GNNs) to predict material properties with near density functional theory (DFT) accuracy. While large-scale pretrained foundation models offer transferable baseline representations, they frequently struggle to generalise to out-of-distribution (OOD) target systems -- a common challenge in modelling complex or chemically diverse materials. Fine-tuning is the standard remedy, but the high cost of generating DFT-labelled configurations confines adaptation to data-scarce regimes, where over-parameterised GNNs amplify overfitting and degrade target-domain performance. To address this, we propose a prototype-based alignment approach for data-efficient fine-tuning of MLIPs. Our method identifies local structural similarities between the source and target domains by grouping atoms with analogous chemical environments based on their latent representations. Each target-domain atom's energy contribution is aligned to its source-domain prototype, introducing an inductive bias that anchors fine-tuned representations to the pretrained structure, encouraging effective reuse of learned interactions and improving generalisation without restrictive assumptions on the target chemistry. We evaluate our method on the rMD17 benchmark using equivariant MACE and invariant SchNet across varying data budgets, and extend evaluation to the MACE-OFF foundation models on the SPICE dataset. Our approach consistently improves predictive accuracy in the low-data regime, reducing energy MAE by up to 18% over standard fine-tuning baselines.

cond-mat.mtrl-sci↗
arXiv2026

The sharp refined Bohr inequalities for a subclass of close-to-convex harmonic mappings

Let $\mathcal{H}$ be the class of normalized complex valued harmonic functions $ f = h + \overline{g}$ defined on the unit disk $\mathbb{D}$, where $h$ and $g$ are analytic functions with the normalization conditions $h(0) = h'(0) - 1 = 0$ and $g(0) = 0$. For the class $R_H^{0}(\gamma, \delta, \lambda)$ ( $0 \leq \lambda < \gamma \leq \delta$) consisting of functions \( f = h+\bar{g} \in \mathcal{H}\) satisfying the condition $f_{\overline{z}}(0)=0$ and the inequality $ Re(\gamma h'(z)+\delta z h''(z) +(\frac{\delta - \gamma}{2})z^2 h'''(z)-\lambda)> |\gamma g'(z)+\delta z g''(z) +(\frac{\delta - \gamma}{2})z^2 g'''(z)|$, we obtain sharp improved Bohr Phenomenon, refined Bohr radius and the Bohr-Rogosinski inequality for the class $R_H^{0}(\gamma, \delta, \lambda)$.

math.CV↗
arXiv2026

LatentMimic: Terrain-Adaptive Locomotion via Latent Space Imitation

Developing natural and diverse locomotion controllers for quadruped robots that can adapt to complex terrains while preserving motion style remains a significant challenge. Existing imitation-based methods face a fundamental optimization trade-off: strict adherence to motion capture (mocap) references penalizes the geometric deviations required for terrain adaptability, whereas terrain-centric policies often compromise stylistic fidelity. We introduce LatentMimic, a novel locomotion learning framework that decouples stylistic fidelity from geometric constraints. By minimizing the marginal latent divergence between the policy's state-action distribution and a learned mocap prior, our approach provides a conditional relaxation of rigid pose-tracking objectives. This formulation preserves gait topology while permitting independent end-effector adaptations for irregular terrains. We further introduce a terrain adaptation module with a dynamic replay buffer to resolve the policy's distribution shifts across different terrains. We validate our method across four locomotion styles and four terrains, demonstrating that LatentMimic enables effective terrain-adaptive locomotion, achieving higher terrain traversal success rates than state-of-the-art motion-tracking methods while maintaining high stylistic fidelity.

cs.RO↗
arXiv2026

Tool-Aware Planning in Contact Center AI: Evaluating LLMs through Lineage-Guided Query Decomposition

We present a domain-grounded framework and benchmark for tool-aware plan generation in contact centers, where answering a query for business insights, our target use case, requires decomposing it into executable steps over structured tools (Text2SQL (T2S)/Snowflake) and unstructured tools (RAG/transcripts) with explicit depends_on for parallelism. Our contributions are threefold: (i) a reference-based plan evaluation framework operating in two modes - a metric-wise evaluator spanning seven dimensions (e.g., tool-prompt alignment, query adherence) and a one-shot evaluator; (ii) a data curation methodology that iteratively refines plans via an evaluator->optimizer loop to produce high-quality plan lineages (ordered plan revisions) while reducing manual effort; and (iii) a large-scale study of 14 LLMs across sizes and families for their ability to decompose queries into step-by-step, executable, and tool-assigned plans, evaluated under prompts with and without lineage. Empirically, LLMs struggle on compound queries and on plans exceeding 4 steps (typically 5-15); the best total metric score reaches 84.8% (Claude-3-7-Sonnet), while the strongest one-shot match rate at the "A+" tier (Extremely Good, Very Good) is only 49.75% (o3-mini). Plan lineage yields mixed gains overall but benefits several top models and improves step executability for many. Our results highlight persistent gaps in tool-understanding, especially in tool-prompt alignment and tool-usage completeness, and show that shorter, simpler plans are markedly easier. The framework and findings provide a reproducible path for assessing and improving agentic planning with tools for answering data-analysis queries in contact-center settings.

cs.CL↗
arXiv2026

Counterfactual Fairness Evaluation of LLM-Based Contact Center Agent Quality Assurance System

Large Language Models (LLMs) are increasingly deployed in contact-center Quality Assurance (QA) to automate agent performance evaluation and coaching feedback. While LLMs offer unprecedented scalability and speed, their reliance on web-scale training data raises concerns regarding demographic and behavioral biases that may distort workforce assessment. We present a counterfactual fairness evaluation of LLM-based QA systems across 13 dimensions spanning three categories: Identity, Context, and Behavioral Style. Fairness is quantified using the Counterfactual Flip Rate (CFR), the frequency of binary judgment reversals, and the Mean Absolute Score Difference (MASD), the average shift in coaching or confidence scores across counterfactual pairs. Evaluating 18 LLMs on 3,000 real-world contact center transcripts, we find systematic disparities, with CFR ranging from 5.4% to 13.0% and consistent MASD shifts across confidence, positive, and improvement scores. Larger, more strongly aligned models show lower unfairness, though fairness does not track accuracy. Contextual priming of historical performance induces the most severe degradations (CFR up to 16.4%), while implicit linguistic identity cues remain a persistent bias source. Finally, we analyze the efficacy of fairness-aware prompting, finding that explicit instructions yield only modest improvements in evaluative consistency. Our findings underscore the need for standardized fairness auditing pipelines prior to deploying LLMs in high-stakes workforce evaluation.

cs.CL↗
arXiv2026

Hilbert Series and Complete-Intersection Structure of Coulomb Branches for Non-Maximal Nilpotent Orbits of $SL(N)$

We study the Coulomb branches of three-dimensional $\mathcal N=4$ quiver gauge theories of type $T_\rho(SU(N))$ associated with non-maximal nilpotent orbits of $SL(N)$. Using the Hall--Littlewood closed form for Coulomb-branch Hilbert series, together with independent checks from the monopole formula, we compute exact unrefined Hilbert series for all non-maximal partitions $\rho\vdash N$ with $N=4$, and extend the analysis to $N=5,6$. By analyzing the plethystic logarithms of the resulting Hilbert series, we find that in all cases examined the Coulomb branch is a complete intersection. The number of generators and relations follows a uniform pattern governed by the transpose partition $\rho^T$, with exactly $N-1$ relations appearing independently of $\rho$ in these examples. We summarize the results in explicit classification tables and formulate conjectures extending these patterns to arbitrary $N$. Our findings provide strong evidence for a remarkable uniformity in the algebraic structure of Coulomb branches within the $T_\rho(SU(N))$ family at low rank.

hep-th↗
arXiv2025

ExpIDS: A Drift-adaptable Network Intrusion Detection System With Improved Explainability

Despite all the advantages associated with Network Intrusion Detection Systems (NIDSs) that utilize machine learning (ML) models, there is a significant reluctance among cyber security experts to implement these models in real-world production settings. This is primarily because of their opaque nature, meaning it is unclear how and why the models make their decisions. In this work, we design a deep learning-based NIDS, ExpIDS to have high decision tree explanation fidelity, i.e., the predictions of decision tree explanation corresponding to ExpIDS should be as close to ExpIDS's predictions as possible. ExpIDS can also adapt to changes in network traffic distribution (drift). With the help of extensive experiments, we verify that ExpIDS achieves higher decision tree explanation fidelity and a malicious traffic detection performance comparable to state-of-the-art NIDSs for common attacks with varying levels of real-world drift.

cs.CR↗
arXiv2025

A Survey of Transaction Tracing Techniques for Blockchain Systems

With the proliferation of new blockchain-based cryptocurrencies/assets and platforms that make it possible to transact across them, it becomes important to consider not just whether the transfer of coins/assets can be tracked within their respective transaction ledger, but also if they can be tracked as they move across ledgers. This is especially important given that there are documented cases of criminals attempting to use these cross-ledger trades to obscure the flow of their coins/assets. In this paper, we perform a systematic review of the various tracing techniques for blockchain transactions proposed in literature, categorize them using multiple criteria (such as tracing approach and targeted objective) and compare them. Based on the above categorization, we provide insights on the state of blockchain transaction tracing literature and identify the limitations of existing approaches. Finally, we suggest directions for future research in this area based on our analysis.

cs.CR↗
arXiv2025

Task Decoding based on Eye Movements using Synthetic Data Augmentation

Machine learning has been extensively used in various applications related to eye-tracking research. Understanding eye movement is one of the most significant subsets of eye-tracking research that reveals the scanning pattern of an individual. Researchers have thoroughly analyzed eye movement data to understand various eye-tracking applications, such as attention mechanisms, navigational behavior, task understanding, etc. The outcome of traditional machine learning algorithms used for decoding tasks based on eye movement data has received a mixed reaction to Yarbus' claim that it is possible to decode the observer's task from their eye movements. In this paper, to support the hypothesis by Yarbus, we are decoding tasks categories while generating synthetic data samples using well-known Synthetic Data Generators CTGAN and its variations such as CopulaGAN and Gretel AI Synthetic Data generators on available data from an in-person user study. Our results show that augmenting more eye movement data combined with additional synthetically generated improves classification accuracy even with traditional machine learning algorithms. We see a significant improvement in task decoding accuracy from 28.1% using Random Forest to 82% using Inception Time when five times more data is added in addition to the 320 real eye movement dataset sample. Our proposed framework outperforms all the available studies on this dataset because of the use of additional synthetic datasets. We validated our claim with various algorithms and combinations of real and synthetic data to show how decoding accuracy increases with the increase in the augmentation of generated data to real data.

cs.AI↗
arXiv2025

Why Synthetic Isn't Real Yet: A Diagnostic Framework for Contact Center Dialogue Generation

Synthetic data is increasingly critical for contact centers, where privacy constraints and data scarcity limit the availability of real conversations. However, generating synthetic dialogues that are realistic and useful for downstream applications remains challenging. In this work, we benchmark multiple generation strategies guided by structured supervision on call attributes (Intent Summaries, Topic Flows, and Quality Assurance (QA) Forms) across multiple languages. To test downstream utility, we evaluate synthetic transcripts on an automated quality assurance (AutoQA) task, finding that prompts optimized on real transcripts consistently outperform those optimized on synthetic transcripts. These results suggest that current synthetic transcripts fall short in capturing the full realism of real agent-customer interactions. To highlight these downstream gaps, we introduce a diagnostic evaluation framework comprising 17 metrics across four dimensions: (1) Emotional and Sentiment Arcs, (2) Linguistic Complexity, (3) Interaction Style, and (4) Conversational Properties. Our analysis shows that even with structured supervision, current generation strategies exhibit measurable deficiencies in sentiment fidelity, disfluency modeling, behavioral variation, and conversational realism. Together, these results highlight the importance of diagnostic, metric-driven evaluation for synthetic conversation generation intended for downstream applications.

cs.CL↗
arXiv2025

Spot the BlindSpots: Systematic Identification and Quantification of Fine-Grained LLM Biases in Contact Center Summaries

Abstractive summarization is a core application in contact centers, where Large Language Models (LLMs) generate millions of summaries of call transcripts daily. Despite their apparent quality, it remains unclear whether LLMs systematically under- or over-attend to specific aspects of the transcript, potentially introducing biases in the generated summary. While prior work has examined social and positional biases, the specific forms of bias pertinent to contact center operations - which we term Operational Bias - have remained unexplored. To address this gap, we introduce BlindSpot, a framework built upon a taxonomy of 15 operational bias dimensions (e.g., disfluency, speaker, topic) for the identification and quantification of these biases. BlindSpot leverages an LLM as a zero-shot classifier to derive categorical distributions for each bias dimension in a pair of transcript and its summary. The bias is then quantified using two metrics: Fidelity Gap (the JS Divergence between distributions) and Coverage (the percentage of source labels omitted). Using BlindSpot, we conducted an empirical study with 2500 real call transcripts and their summaries generated by 20 LLMs of varying scales and families (e.g., GPT, Llama, Claude). Our analysis reveals that biases are systemic and present across all evaluated models, regardless of size or family.

cs.CL↗
arXiv2025

Entanglement witnesses mediated via axionLike particles

Entanglement is solely a quantum property and it can be extremely helpful to test the physics beyond the Standard Model in tabletop experiments with the advent of future quantum technologies. In this work, we provide an entanglement-based partial positive transpose witness for Yukawa-type potentials in the infrared regime between pairs of neutral/charged particles in a spatial quantum superposition. The entanglement is created by the interaction beyond the Standard Model such as axionlike particle or physics motivated by string theory such as extra dimensions in the context of gravity. We will study the constrained couplings in a few different models along with the decoherence rate to show what parameters can be searched for in near-future entanglement-driven experiments for the search of new physics.

quant-ph↗
arXiv2024

Evaluating The Explainability of State-of-the-Art Deep Learning-based Network Intrusion Detection Systems

State-of-the-art deep learning (DL)-based network intrusion detection systems (NIDSs) offer limited "explainability". For example, how do they make their decisions? Do they suffer from hidden correlations? Prior works have applied eXplainable AI (XAI) techniques to ML-based security systems such as conventional ML classifiers trained on public network intrusion datasets, Android malware detection and malicious PDF file detection. However, those works have not evaluated XAI methods on state-of-the-art DL-based NIDSs and do not use latest XAI tools. In this work, we analyze state-of-the-art DL-based NIDS models using conventional as well as recently proposed XAI techniques through extensive experiments with different attack datasets. Furthermore, we introduce a criteria to evaluate the level of agreement between global- and local-level explanations generated for an NIDS. Using this criteria in addition to other security-focused criteria, we compare the explanations generated across XAI methods. The results show that: (1) the decisions of some DL-based NIDS models can be better explained than other models, (2) XAI explanations generated using different tools are in conflict for most of the NIDS models considered in this work and (3) there are significant differences between XAI methods in terms of some security-focused criteria. Based on our results, we make recommendations on how to achieve a balance between explainability and model detection performance.

cs.CR↗
arXiv2024

Automatic AI controller that can drive with confidence: steering vehicle with uncertainty knowledge

In safety-critical systems that interface with the real world, the role of uncertainty in decision-making is pivotal, particularly in the context of machine learning models. For the secure functioning of Cyber-Physical Systems (CPS), it is imperative to manage such uncertainty adeptly. In this research, we focus on the development of a vehicle's lateral control system using a machine learning framework. Specifically, we employ a Bayesian Neural Network (BNN), a probabilistic learning model, to address uncertainty quantification. This capability allows us to gauge the level of confidence or uncertainty in the model's predictions. The BNN based controller is trained using simulated data gathered from the vehicle traversing a single track and subsequently tested on various other tracks. We want to share two significant results: firstly, the trained model demonstrates the ability to adapt and effectively control the vehicle on multiple similar tracks. Secondly, the quantification of prediction confidence integrated into the controller serves as an early-warning system, signaling when the algorithm lacks confidence in its predictions and is therefore susceptible to failure. By establishing a confidence threshold, we can trigger manual intervention, ensuring that control is relinquished from the algorithm when it operates outside of safe parameters.

cs.LG↗
arXiv2023

Towards Probing Contact Center Large Language Models

Fine-tuning large language models (LLMs) with domain-specific instructions has emerged as an effective method to enhance their domain-specific understanding. Yet, there is limited work that examines the core characteristics acquired during this process. In this study, we benchmark the fundamental characteristics learned by contact-center (CC) specific instruction fine-tuned LLMs with out-of-the-box (OOB) LLMs via probing tasks encompassing conversational, channel, and automatic speech recognition (ASR) properties. We explore different LLM architectures (Flan-T5 and Llama), sizes (3B, 7B, 11B, 13B), and fine-tuning paradigms (full fine-tuning vs PEFT). Our findings reveal remarkable effectiveness of CC-LLMs on the in-domain downstream tasks, with improvement in response acceptability by over 48% compared to OOB-LLMs. Additionally, we compare the performance of OOB-LLMs and CC-LLMs on the widely used SentEval dataset, and assess their capabilities in terms of surface, syntactic, and semantic information through probing tasks. Intriguingly, we note a relatively consistent performance of probing classifiers on the set of probing tasks. Our observations indicate that CC-LLMs, while outperforming their out-of-the-box counterparts, exhibit a tendency to rely less on encoding surface, syntactic, and semantic properties, highlighting the intricate interplay between domain-specific adaptation and probing task performance opening up opportunities to explore behavior of fine-tuned language models in specialized contexts.

cs.CL↗
arXiv2023

Malicious Lateral Movement in 5G Core With Network Slicing And Its Detection

5G networks are susceptible to cyber attacks due to reasons such as implementation issues and vulnerabilities in 3GPP standard specifications. In this work, we propose lateral movement strategies in a 5G Core (5GC) with network slicing enabled, as part of a larger attack campaign by well-resourced adversaries such as APT groups. Further, we present 5GLatte, a system to detect such malicious lateral movement. 5GLatte operates on a host-container access graph built using host/NF container logs collected from the 5GC. Paths inferred from the access graph are scored based on selected filtering criteria and subsequently presented as input to a threshold-based anomaly detection algorithm to reveal malicious lateral movement paths. We evaluate 5GLatte on a dataset containing attack campaigns (based on MITRE ATT&CK and FiGHT frameworks) launched in a 5G test environment which shows that compared to other lateral movement detectors based on state-of-the-art, it can achieve higher true positive rates with similar false positive rates.

cs.CR↗