arXiv subjects

Alok Singh

Publications and source records attributed to Alok Singh.

15 recordsLinked to original sources

arXiv2026

A Formal Hierarchical Architecture for Agentic Orchestration with Stack-Based Execution and Lazy Discovery

The rapid expansion of capabilities in Large Language Model (LLM) agents has exposed a critical architectural bottleneck: when agents are given access to a flat, monolithic registry of tools, the model must evaluate hundreds or thousands of options simultaneously. This leads to decision-space explosion, context window saturation, and degraded routing accuracy. To address these limitations, this paper presents a hierarchical, skill-based architecture for agentic orchestration. Capabilities are organized as a rooted tree where internal nodes make routing decisions and leaf nodes execute deterministic tasks. The runtime enforces a single-step execution loop governed by a Last-In-First-Out (LIFO) stack, giving the agent a form of memory akin to a Pushdown Automaton, therefore enabling it to track nested execution contexts and resume deterministically from any depth. Capability discovery follows a manifest-driven, lazy-loading protocol: only the immediate children of the active node are loaded, so memory and prompt costs scale with the explored path rather than the global registry. By replacing global memory with localized stack frames, the architecture prevents outputs from one execution branch from leaking into another, establishing the isolation guarantees required for deployment in regulated enterprise environments. We also discuss UPI Help, an AI-powered digital payments support product, as a motivating production deployment context. We provide a mathematical formalization of the orchestration state, detailed algorithmic analysis of the execution loop, and controlled benchmarks comparing flat and hierarchical routing under increasing tool catalogs, multi-step workflow pressure, and visible schema-token exposure per LLM call.

cs.AI↗

arXiv2025

A benchmark for vericoding: formally verified program synthesis

We present and test the largest benchmark for vericoding, LLM-generation of formally verified code from formal specifications - in contrast to vibe coding, which generates potentially buggy code from a natural language description. Our benchmark contains 12,504 formal specifications, with 3,029 in Dafny, 2,334 in Verus/Rust and 7,141 in Lean. Of these, 6,174 are new unseen problems. We find vericoding success rates of 27% in Lean, 44% in Verus/Rust and 82% in Dafny using off-the-shelf LLMs. Adding natural-language descriptions does not significantly improve performance. We also find that LLM progress has improved progress on pure Dafny verification from 68% to 96% over the past year. The benchmark and vericoding results are shared at https://github.com/Beneficial-AI-Foundation/vericoding-benchmark

cs.SE↗

arXiv2024

DNN-GDITD: Out-of-distribution detection via Deep Neural Network based Gaussian Descriptor for Imbalanced Tabular Data

Classification tasks present challenges due to class imbalances and evolving data distributions. Addressing these issues requires a robust method to handle imbalances while effectively detecting out-of-distribution (OOD) samples not encountered during training. This study introduces a novel OOD detection algorithm designed for tabular datasets, titled Deep Neural Network-based Gaussian Descriptor for Imbalanced Tabular Data (DNN-GDITD). The DNN-GDITD algorithm can be placed on top of any DNN to facilitate better classification of imbalanced data and OOD detection using spherical decision boundaries. Using a combination of Push, Score-based, and focal losses, DNN-GDITD assigns confidence scores to test data points, categorizing them as known classes or as an OOD sample. Extensive experimentation on tabular datasets demonstrates the effectiveness of DNN-GDITD compared to three OOD algorithms. Evaluation encompasses imbalanced and balanced scenarios on diverse tabular datasets, including a synthetic financial dispute dataset and publicly available tabular datasets like Gas Sensor, Drive Diagnosis, and MNIST, showcasing DNN-GDITD's versatility.

cs.LG↗

arXiv2022

Entity Alignment For Knowledge Graphs: Progress, Challenges, and Empirical Studies

Entity Alignment (EA) identifies entities across databases that refer to the same entity. Knowledge graph-based embedding methods have recently dominated EA techniques. Such methods map entities to a low-dimension space and align them based on their similarities. With the corpus of EA methodologies growing rapidly, this paper presents a comprehensive analysis of various existing EA methods, elaborating their applications and limitations. Further, we distinguish the methods based on their underlying algorithms and the information they incorporate to learn entity representations. Based on challenges in industrial datasets, we bring forward $4$ research questions (RQs). These RQs empirically analyse the algorithms from the perspective of \textit{Hubness, Degree distribution, Non-isomorphic neighbourhood,} and \textit{Name bias}. For Hubness, where one entity turns up as the nearest neighbour of many other entities, we define an $h$-score to quantify its effect on the performance of various algorithms. Additionally, we try to level the playing field for algorithms that rely primarily on name-bias existing in the benchmarking open-source datasets by creating a low name bias dataset. We further create an open-source repository for $14$ embedding-based EA methods and present the analysis for invoking further research motivations in the field of EA.

cs.AI↗

arXiv2021

A Reinforcement Learning Environment for Mathematical Reasoning via Program Synthesis

We convert the DeepMind Mathematics Dataset into a reinforcement learning environment by interpreting it as a program synthesis problem. Each action taken in the environment adds an operator or an input into a discrete compute graph. Graphs which compute correct answers yield positive reward, enabling the optimization of a policy to construct compute graphs conditioned on problem statements. Baseline models are trained using Double DQN on various subsets of problem types, demonstrating the capability to learn to correctly construct graphs despite the challenges of combinatorial explosion and noisy rewards.

cs.LG↗

arXiv2020

A Comprehensive Review on Recent Methods and Challenges of Video Description

Video description involves the generation of the natural language description of actions, events, and objects in the video. There are various applications of video description by filling the gap between languages and vision for visually impaired people, generating automatic title suggestion based on content, browsing of the video based on the content and video-guided machine translation [86] etc.In the past decade, several works had been done in this field in terms of approaches/methods for video description, evaluation metrics,and datasets. For analyzing the progress in the video description task, a comprehensive survey is needed that covers all the phases of video description approaches with a special focus on recent deep learning approaches. In this work, we report a comprehensive survey on the phases of video description approaches, the dataset for video description, evaluation metrics, open competitions for motivating the research on the video description, open challenges in this field, and future research directions. In this survey, we cover the state-of-the-art approaches proposed for each and every dataset with their pros and cons. For the growth of this research domain,the availability of numerous benchmark dataset is a basic need. Further, we categorize all the dataset into two classes: open domain dataset and domain-specific dataset. From our survey, we observe that the work in this field is in fast-paced development since the task of video description falls in the intersection of computer vision and natural language processing. But still, the work in the video description is far from saturation stage due to various challenges like the redundancy due to similar frames which affect the quality of visual features, the availability of dataset containing more diverse content and availability of an effective evaluation metric.

cs.CV↗

arXiv2020

NITS-VC System for VATEX Video Captioning Challenge 2020

Video captioning is process of summarising the content, event and action of the video into a short textual form which can be helpful in many research areas such as video guided machine translation, video sentiment analysis and providing aid to needy individual. In this paper, a system description of the framework used for VATEX-2020 video captioning challenge is presented. We employ an encoder-decoder based approach in which the visual features of the video are encoded using 3D convolutional neural network (C3D) and in the decoding phase two Long Short Term Memory (LSTM) recurrent networks are used in which visual features and input captions are fused separately and final output is generated by performing element-wise product between the output of both LSTMs. Our model is able to achieve BLEU scores of 0.20 and 0.22 on public and private test data sets respectively.

cs.CV↗

arXiv2019

Detecting Spiky Corruption in Markov Decision Processes

Current reinforcement learning methods fail if the reward function is imperfect, i.e. if the agent observes reward different from what it actually receives. We study this problem within the formalism of Corrupt Reward Markov Decision Processes (CRMDPs). We show that if the reward corruption in a CRMDP is sufficiently "spiky", the environment is solvable. We fully characterize the regret bound of a Spiky CRMDP, and introduce an algorithm that is able to detect its corrupt states. We show that this algorithm can be used to learn the optimal policy with any common reinforcement learning algorithm. Finally, we investigate our algorithm in a pair of simple gridworld environments, finding that our algorithm can detect the corrupt states and learn the optimal policy despite the corruption.

cs.LG↗

arXiv2019

Towards A Methodology and Framework for Workflow-Driven Team Science

Scientific workflows are powerful tools for management of scalable experiments, often composed of complex tasks running on distributed resources. Existing cyberinfrastructure provides components that can be utilized within repeatable workflows. However, data and computing advances continuously change the way scientific workflows get developed and executed, pushing the scientific activity to be more data-driven, heterogeneous and collaborative. Workflow development today depends on the effective collaboration and communication of a cross-disciplinary team, not only with humans but also with analytical systems and infrastructure. This paper presents a collaboration-centered reference architecture to extend workflow systems with dynamic, predictable and programmable interfaces to systems and infrastructure while bridging the exploratory and scalable activities in the scientific process. We also present a conceptual design towards the development of methodologies and tools for effective workflow-driven collaborations, namely the PPoDS methodology and the SmartFlows Toolkit for smart utilization of workflows in a rapidly evolving cyberinfrastructure ecosystem.

cs.CY↗

arXiv2018

Deep Learning on Operational Facility Data Related to Large-Scale Distributed Area Scientific Workflows

Distributed computing platforms provide a robust mechanism to perform large-scale computations by splitting the task and data among multiple locations, possibly located thousands of miles apart geographically. Although such distribution of resources can lead to benefits, it also comes with its associated problems such as rampant duplication of file transfers increasing congestion, long job completion times, unexpected site crashing, suboptimal data transfer rates, unpredictable reliability in a time range, and suboptimal usage of storage elements. In addition, each sub-system becomes a potential failure node that can trigger system wide disruptions. In this vision paper, we outline our approach to leveraging Deep Learning algorithms to discover solutions to unique problems that arise in a system with computational infrastructure that is spread over a wide area. The presented vision, motivated by a real scientific use case from Belle II experiments, is to develop multilayer neural networks to tackle forecasting, anomaly detection and optimization challenges in a complex and distributed data movement environment. Through this vision based on Deep Learning principles, we aim to achieve reduced congestion events, faster file transfer rates, and enhanced site reliability.

cs.DC↗

arXiv2018

Modular Resource Centric Learning for Workflow Performance Prediction

Workflows provide an expressive programming model for fine-grained control of large-scale applications in distributed computing environments. Accurate estimates of complex workflow execution metrics on large-scale machines have several key advantages. The performance of scheduling algorithms that rely on estimates of execution metrics degrades when the accuracy of predicted execution metrics decreases. This in-progress paper presents a technique being developed to improve the accuracy of predicted performance metrics of large-scale workflows on distributed platforms. The central idea of this work is to train resource-centric machine learning agents to capture complex relationships between a set of program instructions and their performance metrics when executed on a specific resource. This resource-centric view of a workflow exploits the fact that predicting execution times of sub-modules of a workflow requires monitoring and modeling of a few dynamic and static features. We transform the input workflow that is essentially a directed acyclic graph of actions into a Physical Resource Execution Plan (PREP). This transformation enables us to model an arbitrarily complex workflow as a set of simpler programs running on physical nodes. We delegate a machine learning model to capture performance metrics for each resource type when it executes different program instructions under varying degrees of resource contention. Our algorithm takes the prediction metrics from each resource agent and composes the overall workflow performance metrics by utilizing the structure of the corresponding Physical Resource Execution Plan.

cs.DC↗

arXiv2012

The effect of precipitation on strength and ductility in a Mg-Zn-Y alloy

The effect of pre-ageing deformation on the size and distribution of beta-prime precipitates and subsequently on the resulting strength and ductility have been measured in a Mg-3.0at.%Zn-0.5at.%Y alloy. The alloy was extruded and then subjected to a T8 heat treatment comprised of a solution-treatment, cold-work and artificial ageing. Extrusion was used to introduce texture, ensuring that deformation occurred via slip rather than twinning. Samples were subjected to controlled uniaxial deformation and then isothermally aged to peak hardness. Precipitate length, diameter and number density were measured and evaluated in terms of the strength and ductility of the alloy. The nucleation of the beta-prime precipitates in peak-aged condition without pre-ageing deformation (i.e.T6 treatment) was poor, with only 0.5% volume fraction, compared to approximately 3.5% in T6 treated binary Mg-3.0at.%Zn alloy. The microstructure of the Mg-Zn-Y alloy was less refined, with larger diameter precipitates and lower beta-prime number densities compared to a binary Mg-3.0at.%Zn alloy. Deformation to 5% plastic strain increased the volume fraction of beta-prime precipitates to approximately 2.3% and refined the beta-prime precipitate length and diameter. The combination of these effects increased the yield strength after isothermal ageing from 217MPa (0% cold-work) to 287 MPa (5% cold-work). The yield stress increased linearly with reciprocal interparticle spacing on the basal and prismatic planes and the alloy showed similar strengthening against basal slip to Mg-Zn. The elongation increased linearly with particle spacing. The ductility of Mg-Zn-Y alloys was similar to that of Mg-Zn for equivalently spaced particles.

cond-mat.mtrl-sci↗

arXiv2012

The effect of size and distribution of rod-shaped β' precipitates on the strength and ductility of a Mg-Zn alloy

We report on a quantitative investigation into the effect of size and distribution of rod-shaped β' precipitates on strength and ductility of a Mg-Zn alloy. Despite precipitation strengthening being crucial for the practical application of magnesium alloys this study represents the first systematic examination of the effect of controlled deformation on the precipitate size distribution and the resulting strength and ductility of a magnesium alloy. Pre-ageing deformation was used to obtain various distributions of rod-shaped β' precipitates through heterogeneous nucleation. Alloys were extruded to obtain a texture so as to avoid formation of twins and thus to ensure that dislocations were the primary nucleation site. Pre-ageing strain refined precipitate length and diameter, with average length reduced from 440 nm to 60 nm and diameter from 14 nm to 9 nm. Interparticle spacings were measured from micrographs and indicated some inhomogeneity in the precipitate distribution. The yield stress of the alloy increased from 273 MPa to 309 MPa. The yield stress increased linearly as a function of reciprocal interparticle spacing, but at a lower rate than predicted for Orowan strengthening. Pre-ageing deformation also resulted in a significant loss of ductility (from 17% to 6% elongation). Both true strain at failure and uniform elongation showed a linear relationship with particle spacing, in agreement with models for the accumulation of dislocations around non-deforming obstacles. Samples subjected to 3% pre-ageing deformation showed a substantially increased ageing response compared to non-deformed material; however, additional deformation (to 5% strain) resulted in only modest changes in precipitate distribution and mechanical properties.

cond-mat.mtrl-sci↗

arXiv2012

Structural relationships among monoclinic and hexagonal phases and transition structures in Mg-Zn-Y alloys

Isothermal ageing of plastically deformed Mg-Zn-Y alloys resulted in precipitation along {10-12} twin boundaries. The bulky precipitates formed had structures similar to those recently reported for the rod-like β' precipitates, but afforded a more detailed study by high resolution TEM due to their larger size. The core of the precipitates often had the structure of monoclinic Mg4Zn7 phase, and had the orientation [0001]_Mg//[010]; {10-10}_Mg // (201) with either the matrix or the twin. On this Mg4Zn7 phase, hexagonal MgZn2 phase grew in two orientations, both with [010]_Mg4Zn7 // [11-20]_MgZn2. One of these orientations formed a known orientation relationship [0001]_Mg // [11-20]_MgZn2; {11-20}_Mg // (0001)_MgZn2 with the matrix. The part of the precipitate with the MgZn2 structure was usually in direct contact with the twin boundary. Both Mg4Zn7 and MgZn2 phases have layered structures that can be described with similar building blocks of icosahedrally coordinated atoms. The atomic positions of zinc atoms comprise the vertices of these icosahedra and form "thick" rhombic tiles. Orientation of these rhombuses remain unchanged across the interfaces between the two phases. Near the interface with MgZn2, transition structures formed in Mg4Zn7 phase, with Zn:Mg atom ratio between those of Mg4Zn7 and MgZn2 phases. In these transition structures, the unit cell of Mg4Zn7 phase is extended along [100] or [001] by half a unit cell length by continuation of the rhombic tiling. Structures of these extended unit cells are proposed.

cond-mat.mtrl-sci↗

arXiv2012

Orientation relationships between icosahedral clusters in hexagonal MgZn2 and monoclinic Mg4Zn7 phases in Mg-Zn(-Y) alloys

Intermetallic precipitates formed in heat-treated and aged Mg-Zn and Mg-Zn-Y alloys have been investigated via electron microscopy. Coarse spheroidal precipitates formed on deformation twin boundaries contained domains belonging to either the MgZn2 hexagonal Laves phase or the monoclinic Mg4Zn7 phase. Both phases are structurally related to the quasi-crystalline phase formed in Mg-Zn-Y alloys, containing icosahedrally coordinated zinc atoms arranged as a series of broad rhombohedral units. This rhombohedral arrangement was also visible in intragranular precipitates where local regions with the structures of hexagonal MgZn2 and Mg4Zn7 were found. The orientation adopted by the MgZn2 and Mg4Zn7 phases in twin-boundary and intragranular precipitates was such that the icosahedral clusters were aligned similarly. These results highlight the close structural similarities between the precipitates of the Mg-Zn-Y alloy system.

cond-mat.mtrl-sci↗