arXiv subjects

Parth Shah

Publications and source records attributed to Parth Shah.

At least 19 recordsLinked to original sources

arXiv2026

NAAMSE: Framework for Evolutionary Security Evaluation of Agents

AI agents are increasingly deployed in production, yet their security evaluations remain bottlenecked by manual red-teaming or static benchmarks that fail to model adaptive, multi-turn adversaries. We propose NAAMSE, an evolutionary framework that reframes agent security evaluation as a feedback-driven optimization problem. Our system employs a single autonomous agent that orchestrates a lifecycle of genetic prompt mutation, hierarchical corpus exploration, and asymmetric behavioral scoring. By using model responses as a fitness signal, the framework iteratively compounds effective attack strategies while simultaneously ensuring "benign-use correctness", preventing the degenerate security of blanket refusal. Our experiments across a diverse suite of state-of-the-art large language models demonstrate that evolutionary mutation systematically amplifies vulnerabilities missed by one-shot methods, with controlled ablations revealing that the synergy between exploration and targeted mutation uncovers high-severity failure modes. We show that this adaptive approach provides a more realistic and scalable assessment of agent robustness in the face of evolving threats. The code for NAAMSE is open source and available at https://github.com/HASHIRU-AI/NAAMSE.

cs.AI↗

arXiv2025

How Many Instructions Can LLMs Follow at Once?

Production-grade LLM systems require robust adherence to dozens or even hundreds of instructions simultaneously. However, the instruction-following capabilities of LLMs at high instruction densities have not yet been characterized, as existing benchmarks only evaluate models on tasks with a single or few instructions. We introduce IFScale, a simple benchmark of 500 keyword-inclusion instructions for a business report writing task to measure how instruction-following performance degrades as instruction density increases. We evaluate 20 state-of-the-art models across seven major providers and find that even the best frontier models only achieve 68% accuracy at the max density of 500 instructions. Our analysis reveals model size and reasoning capability to correlate with 3 distinct performance degradation patterns, bias towards earlier instructions, and distinct categories of instruction-following errors. Our insights can help inform design of instruction-dense prompts in real-world applications and highlight important performance-latency tradeoffs. We open-source the benchmark and all results for further analysis at https://distylai.github.io/IFScale.

cs.AI↗

arXiv2025

HASHIRU: Hierarchical Agent System for Hybrid Intelligent Resource Utilization

Rapid Large Language Model (LLM) advancements are fueling autonomous Multi-Agent System (MAS) development. However, current frameworks often lack flexibility, resource awareness, model diversity, and autonomous tool creation. This paper introduces HASHIRU (Hierarchical Agent System for Hybrid Intelligent Resource Utilization), a novel MAS framework enhancing flexibility, resource efficiency, and adaptability. HASHIRU features a "CEO" agent dynamically managing specialized "employee" agents, instantiated based on task needs and resource constraints (cost, memory). Its hybrid intelligence prioritizes smaller, local LLMs (via Ollama) while flexibly using external APIs and larger models when necessary. An economic model with hiring/firing costs promotes team stability and efficient resource allocation. The system also includes autonomous API tool creation and a memory function. Evaluations on tasks like academic paper review (58% success), safety assessments (100% on a JailbreakBench subset), and complex reasoning (outperforming Gemini 2.0 Flash on GSM8K: 96% vs. 61%; JEEBench: 80% vs. 68.3%; SVAMP: 92% vs. 84%) demonstrate HASHIRU's capabilities. Case studies illustrate its self-improvement via autonomous cost model generation, tool integration, and budget management. HASHIRU offers a promising approach for more robust, efficient, and adaptable MAS through dynamic hierarchical control, resource-aware hybrid intelligence, and autonomous functional extension. Source code and benchmarks are available at https://github.com/HASHIRU-AI/HASHIRU and https://github.com/HASHIRU-AI/HASHIRUBench respectively, and a live demo is available at https://hashiruagentx-hashiruai.hf.space upon request.

cs.MA↗

arXiv2024

On Adversarial Robustness and Out-of-Distribution Robustness of Large Language Models

The increasing reliance on large language models (LLMs) for diverse applications necessitates a thorough understanding of their robustness to adversarial perturbations and out-of-distribution (OOD) inputs. In this study, we investigate the correlation between adversarial robustness and OOD robustness in LLMs, addressing a critical gap in robustness evaluation. By applying methods originally designed to improve one robustness type across both contexts, we analyze their performance on adversarial and out-of-distribution benchmark datasets. The input of the model consists of text samples, with the output prediction evaluated in terms of accuracy, precision, recall, and F1 scores in various natural language inference tasks. Our findings highlight nuanced interactions between adversarial robustness and OOD robustness, with results indicating limited transferability between the two robustness types. Through targeted ablations, we evaluate how these correlations evolve with different model sizes and architectures, uncovering model-specific trends: smaller models like LLaMA2-7b exhibit neutral correlations, larger models like LLaMA2-13b show negative correlations, and Mixtral demonstrates positive correlations, potentially due to domain-specific alignment. These results underscore the importance of hybrid robustness frameworks that integrate adversarial and OOD strategies tailored to specific models and domains. Further research is needed to evaluate these interactions across larger models and varied architectures, offering a pathway to more reliable and generalizable LLMs.

cs.CL↗

arXiv2024

Generic autonomous system approach to interacting dark energy models

We explore an autonomous system analysis of dark energy models with interactions between dark energy and cold dark matter in a general systematic approach to cosmological fluids. We investigate two types of models such as local and non-local ones. In particular, a local form of interaction is directly proportional to only the energy density, while a non-local interaction is directly proportional to the energy density as well as the Hubble parameter. As a consequence, it is explicitly demonstrated that in both cases there exist the stability points in terms of cosmological parameters. This work aims at obtaining acceleration and stability using interaction models without modifying the matter or geometric component of the Universe.

gr-qc↗

arXiv2024

Qualitative behaviour of higher-curvature gravity with boundary terms i.e the f(Q) gravity models by dynamical system analysis

The higher-curvature gravity with boundary terms i.e the $f(Q)$ theories, grounded on non-metricity as a fundamental geometric quantity, exhibit remarkable efficacy in portraying late-time universe phenomena. The aim is to delineate constraints on two prevalent models within this framework, namely the Log-square-root model and the Hyperbolic tangent-power model, by employing the framework of Big Bang Nucleosynthesis (BBN). The approach involves elucidating deviations induced by higher-curvature gravity with boundary terms in the freeze-out temperature ($T_{f}$) concerning its departure from the standard $\Lambda$CDM evolution. Subsequently, constraints on pertinent model parameters are established by imposing limitations on $\vert \frac{\delta T_{f}}{T_{f}}\vert$ derived from observational bounds. This investigation employs dynamical system analysis, scrutinizing both background and perturbed equations. The study systematically explores the phase space of the models, identifying equilibrium points, evaluating their stability, and comprehending the system's trajectory around each critical point. The principal findings of this analysis reveal the presence of a matter-dominated saddle point characterized by the appropriate matter perturbation growth rate. Subsequently, this phase transitions into a stable phase of a dark-energy-dominated, accelerating universe, marked by consistent matter perturbations. Overall, the study substantiates observational confrontations, affirming the potential of higher-curvature gravity with boundary terms as a promising alternative to the $\Lambda$CDM concordance model. The methodological approach underscores the significance of dynamical systems as an independent means to validate and comprehend the cosmological implications of these theories.

gr-qc↗

arXiv2024

Stability analysis of f(Q) gravity models using dynamical systems

In recent years, the modified theory of gravity known as $f(Q)$ gravity has drawn interest as a potential alternative to general relativity. According to this theory, the gravitational force is determined by a function of the so-called ``non-metricity" tensor $Q$, which expresses how far a particle space-time is from the metric geometry. In contrast to general relativity, which describes the gravitational field using the curvature tensor, $f(Q)$ gravity builds a theory of gravity using the non-metricity tensor. For this class of theories, dynamical system analysis of the background and perturbation equations has been carried out in this work to determine how various models behave cosmologically. Here, the critical points are determined for two $f(Q)$ models from the literature: the power law, $ f(Q)=Q+mQ^{n} $, and the logarithmic, $ f(Q)=\alpha+\beta log Q $ models. The stability behaviour and corresponding cosmology are displayed for each critical point. For the power law model, we achieve a matter-dominated saddle point with the right matter perturbation growth rate. For the logarithmic model, we get a saddle point dominated by the geometric component of the $ f(Q)$ model with perturbations in the decomposition of matter. For both models, we later achieved a stable and accelerating Universe with constant matter perturbations.

gr-qc↗

arXiv2024

Autonomous systems and attractor behaviors in non-metricity gravity: stability analysis and cosmic acceleration

The cosmological dynamics are rigorously investigated through the systematic application of autonomous system analysis to the gravitational field equations in non-metricity gravity. The systematic procedure to analyze the late-time cosmic acceleration in higher-order non-metricity gravity is demonstrated by exploring non-hyperbolic critical points with the center manifold theory. The stability properties of these critical points are also evaluated based on the analysis of eigenvalues and phase portraits. It is explicitly shown that the stable node can be realized. The critical points of each model are individually analyzed, and their corresponding cosmological implications are derived. The stability properties of these critical points are evaluated based on the analysis of eigenvalues and phase portraits, revealing that each model includes at least one stable node. Furthermore, the evolution plots of the cosmological parameters confirm the models capacity to exhibit accelerated expansion.

gr-qc↗

arXiv2022

Adaptive Fine-Grained Sketch-Based Image Retrieval

The recent focus on Fine-Grained Sketch-Based Image Retrieval (FG-SBIR) has shifted towards generalising a model to new categories without any training data from them. In real-world applications, however, a trained FG-SBIR model is often applied to both new categories and different human sketchers, i.e., different drawing styles. Although this complicates the generalisation problem, fortunately, a handful of examples are typically available, enabling the model to adapt to the new category/style. In this paper, we offer a novel perspective -- instead of asking for a model that generalises, we advocate for one that quickly adapts, with just very few samples during testing (in a few-shot manner). To solve this new problem, we introduce a novel model-agnostic meta-learning (MAML) based framework with several key modifications: (1) As a retrieval task with a margin-based contrastive loss, we simplify the MAML training in the inner loop to make it more stable and tractable. (2) The margin in our contrastive loss is also meta-learned with the rest of the model. (3) Three additional regularisation losses are introduced in the outer loop, to make the meta-learned FG-SBIR model more effective for category/style adaptation. Extensive experiments on public datasets suggest a large gain over generalisation and zero-shot based approaches, and a few strong few-shot baselines.

cs.CV↗

arXiv2021

Cosmological dynamics of f(R) models in dynamical system analysis

In this work we try to understand the late time acceleration of the universe by assuming some modification in the geometry of the space and using dynamical system analysis. This technique allows to understand the behavior of the universe without analytically solving the field equations. We study the acceleration phase of the universe and stability properties of the critical points which could be compared with observational results. We consider an asymptotic behavior of two particular models $f(R) = R - \mu R_{c} \frac{(R/R_c)^{2n}}{(R/R_c)^{2n} + 1}$ and $f(R) = R - \mu R_{c} \left[ 1 - (1 + R^2/R_{c}^2)^{-n} \right]$ with $n,\mu, R_c >0$ for the study. As a first case we fix the value of $\mu$ and analyzed for all $n$. Later as second case, we fix the value of $n$ and calculation are done for all $\mu$. At the end all the calculations for the generalized case have been shown and results have been discussed in detail.

gr-qc↗

arXiv2020

Regex Queries over Incomplete Knowledge Bases

We propose the novel task of answering regular expression queries (containing disjunction ($\vee$) and Kleene plus ($+$) operators) over incomplete KBs. The answer set of these queries potentially has a large number of entities, hence previous works for single-hop queries in KBC that model a query as a point in high-dimensional space are not as effective. In response, we develop RotatE-Box -- a novel combination of RotatE and box embeddings. It can model more relational inference patterns compared to existing embedding based models. Furthermore, we define baseline approaches for embedding based KBC models to handle regex operators. We demonstrate performance of RotatE-Box on two new regex-query datasets introduced in this paper, including one where the queries are harvested based on actual user query logs. We find that our final RotatE-Box model significantly outperforms models based on just RotatE and just box embeddings.

cs.CL↗

arXiv2020

Neural Machine Translation System of Indic Languages -- An Attention based Approach

Neural machine translation (NMT) is a recent and effective technique which led to remarkable improvements in comparison of conventional machine translation techniques. Proposed neural machine translation model developed for the Gujarati language contains encoder-decoder with attention mechanism. In India, almost all the languages are originated from their ancestral language - Sanskrit. They are having inevitable similarities including lexical and named entity similarity. Translating into Indic languages is always be a challenging task. In this paper, we have presented the neural machine translation system (NMT) that can efficiently translate Indic languages like Hindi and Gujarati that together covers more than 58.49 percentage of total speakers in the country. We have compared the performance of our NMT model with automatic evaluation matrices such as BLEU, perplexity and TER matrix. The comparison of our network with Google translate is also presented where it outperformed with a margin of 6 BLEU score on English-Gujarati translation.

cs.CL↗

arXiv2019

Stability analysis for cosmological models in $f(R)$ gravity using dynamical system analysis

Modified gravity theories have received increased attention lately to understand the late time acceleration of the universe. This viewpoint essentially modifies the geometric components of the universe. Among numerous extension to Einstein's theory of gravity, theories which include higher order curvature invariant, and specifically the class of $f(R)$ theories, have received several acknowledgments. In our current work we try to understand the late time acceleration of the universe by modifying the geometry of the space and using dynamical system analysis. The use of this technique allows to understand the behavior of the universe under several circumstances. Apart from that we study the stability properties of the critical point and acceleration phase of the universe which could then be analyzed with observational data. We consider a particular model $f(R) = R - \mu R_{c}(R/R_{c})^{p}$ with $ 0 < p < 1, \mu, R_{c} > 0$ for the study. As a first case we consider the matter and radiation component of the universe with an assumption of no interaction between them. Later, as a second case we take matter, radiation and dark energy (cosmological constant) where study on effects of linear, non-linear and no interaction between matter and dark energy is considered and results have been discussed in detail.

gr-qc↗

arXiv2019

Nonlinear Semi-Parametric Models for Survival Analysis

Semi-parametric survival analysis methods like the Cox Proportional Hazards (CPH) regression (Cox, 1972) are a popular approach for survival analysis. These methods involve fitting of the log-proportional hazard as a function of the covariates and are convenient as they do not require estimation of the baseline hazard rate. Recent approaches have involved learning non-linear representations of the input covariates and demonstrate improved performance. In this paper we argue against such deep parameterizations for survival analysis and experimentally demonstrate that more interpretable semi-parametric models inspired from mixtures of experts perform equally well or in some cases better than such overly parameterized deep models.

cs.LG↗

arXiv2019

Optimal Approach for Image Recognition using Deep Convolutional Architecture

In the recent time deep learning has achieved huge popularity due to its performance in various machine learning algorithms. Deep learning as hierarchical or structured learning attempts to model high level abstractions in data by using a group of processing layers. The foundation of deep learning architectures is inspired by the understanding of information processing and neural responses in human brain. The architectures are created by stacking multiple linear or non-linear operations. The article mainly focuses on the state-of-art deep learning models and various real world applications specific training methods. Selecting optimal architecture for specific problem is a challenging task, at a closing stage of the article we proposed optimal approach to deep convolutional architecture for the application of image recognition.

cs.CV↗

arXiv2018

Making Sense of Vision and Touch: Self-Supervised Learning of Multimodal Representations for Contact-Rich Tasks

Contact-rich manipulation tasks in unstructured environments often require both haptic and visual feedback. However, it is non-trivial to manually design a robot controller that combines modalities with very different characteristics. While deep reinforcement learning has shown success in learning control policies for high-dimensional inputs, these algorithms are generally intractable to deploy on real robots due to sample complexity. We use self-supervision to learn a compact and multimodal representation of our sensory inputs, which can then be used to improve the sample efficiency of our policy learning. We evaluate our method on a peg insertion task, generalizing over different geometry, configurations, and clearances, while being robust to external perturbations. Results for simulated and real robot experiments are presented.

cs.RO↗

arXiv2018

Improved Parallel Rabin-Karp Algorithm Using Compute Unified Device Architecture

String matching algorithms are among one of the most widely used algorithms in computer science. Traditional string matching algorithms efficiency of underlaying string matching algorithm will greatly increase the efficiency of any application. In recent years, Graphics processing units are emerged as highly parallel processor. They out perform best of the central processing units in scientific computation power. By combining recent advancement in graphics processing units with string matching algorithms will allows to speed up process of string matching. In this paper we proposed modified parallel version of Rabin-Karp algorithm using graphics processing unit. Based on that, result of CPU as well as parallel GPU implementations are compared for evaluating effect of varying number of threads, cores, file size as well as pattern size.

cs.DC↗

arXiv2018

Motion-based Object Segmentation based on Dense RGB-D Scene Flow

Given two consecutive RGB-D images, we propose a model that estimates a dense 3D motion field, also known as scene flow. We take advantage of the fact that in robot manipulation scenarios, scenes often consist of a set of rigidly moving objects. Our model jointly estimates (i) the segmentation of the scene into an unknown but finite number of objects, (ii) the motion trajectories of these objects and (iii) the object scene flow. We employ an hourglass, deep neural network architecture. In the encoding stage, the RGB and depth images undergo spatial compression and correlation. In the decoding stage, the model outputs three images containing a per-pixel estimate of the corresponding object center as well as object translation and rotation. This forms the basis for inferring the object segmentation and final object scene flow. To evaluate our model, we generated a new and challenging, large-scale, synthetic dataset that is specifically targeted at robotic manipulation: It contains a large number of scenes with a very diverse set of simultaneously moving 3D objects and is recorded with a simulated, static RGB-D camera. In quantitative experiments, we show that we outperform state-of-the-art scene flow and motion-segmentation methods on this data set. In qualitative experiments, we show how our learned model transfers to challenging real-world scenes, visually generating better results than existing methods.

cs.RO↗