arXiv subjects

Junaid Ahmed

Publications and source records attributed to Junaid Ahmed.

8 recordsLinked to original sources

arXiv2026

Computational Depth Measurement in Thermographic Video: Overcoming Spatial Overfitting via Spatio-Temporal Decoupling

Accurate through-thickness measurement of subsurface delamination depth in Carbon Fiber Reinforced Polymer (CFRP) is important for structural assessment because defect location determines affected load-bearing layers. Optical pulsed thermography (OPT) provides a two-dimensional thermal video rather than volumetric measurements, so depth must be inferred from temporal heat-diffusion responses. A challenge is spatial dataset bias: when calibration defects follow regular grids, regression models may memorize their geometry instead of learning physical relationship between thermal decay and depth. This work introduces a spatio-temporal decoupling architecture that separates spatial defect localization from temporal depth measurement. Defect regions are first localized using segmentation methods, after which thermal responses are spatially averaged and converted into sixteen physics-informed temporal, energy, statistical, and geometric features. These features expose the one-dimensional heat-conduction relationship while withholding pixel coordinates from the depth model. Four regression models are evaluated using specimen-level cross-validation: Random Forest (RF), Gradient Boosting Machine (GBM), Advanced Multi-Layer Perceptron (Adv-MLP), and XGBoost. Unregularized trees and over-parameterized Adv-MLP exhibit calibration collapse under geometric shifts, with errors exceeding 0.5 mm. In contrast, regularized XGBoost with L1/L2 penalties and column sampling maintains cross-specimen calibration, achieving a mean absolute error (MAE) of 0.056 mm and root mean square error (RMSE) of 0.085 mm. Predicted depths are merged with masks to generate Delaunay-triangulated three-dimensional defect models in three to five seconds per specimen. Results show that mathematical regularization and spatio-temporal decoupling reduce spatial memorization in thermal-video depth regression.

cs.AI↗

arXiv2026

AutoResearch: An Execution-Grounded Multi-Agent Framework for Reliable Research Workflow Automation

Automated research agents increasingly generate code, retrieve literature, and draft scientific artifacts, but they often fail to verify whether generated experiments execute correctly or whether cited sources support generated claims. We present AutoResearch, an execution-grounded multi-agent framework for reliable research workflow automation. AutoResearch couples sandboxed Python/PyTorch execution, iterative code repair, citation verification, claim-support auditing, decision control, and structured \LaTeX{} artifact generation. The system treats runtime errors, citation-verification failures, and review-agent feedback as practical filtering signals for generated research artifacts. In controlled evaluations on HumanEval, MBPP, a SciCode subset, citation-validation tasks, claim-support auditing, and small end-to-end workflow stress tests, AutoResearch improves execution success, citation validity, local claim support, and workflow completion relative to directly comparable baselines. Code-oriented agents are reported separately as partial comparisons. AutoResearch is intended as a reliability-oriented research assistant, not as a fully autonomous scientist or a standalone manuscript-quality benchmark. Source Code: https://github.com/raja21068/AutoResearch

cs.CY↗

arXiv2026

AgentForge: Execution-Grounded Multi-Agent LLM Framework for Autonomous Software Engineering

Large language models generate plausible code but cannot verify correctness. Existing multi-agent systems simulate execution or leave verification optional. We introduce execution-grounded verification as a first-class principle: every code change must survive sandboxed execution before propagation. We instantiate this principle in AGENTFORGE, a multi-agent framework where Planner, Coder, Tester, Debugger, and Critic agents coordinate through shared memory and a mandatory Docker sandbox. We formalize software engineering with LLMs as an iterative decision process over repository states, where execution feedback provides a stronger supervision signal than next-token likelihood. AGENTFORGE achieves 40.0\% resolution on SWE-BENCH Lite, outperforming single-agent baselines by 26--28 points. Ablations confirm that execution feedback and role decomposition each independently drive performance. The framework is open-source at https://github.com/raja21068/AutoCodeAI.

cs.SE↗

arXiv2023

Graph Neural Operators for Classification of Spatial Transcriptomics Data

The inception of spatial transcriptomics has allowed improved comprehension of tissue architectures and the disentanglement of complex underlying biological, physiological, and pathological processes through their positional contexts. Recently, these contexts, and by extension the field, have seen much promise and elucidation with the application of graph learning approaches. In particular, neural operators have risen in regards to learning the mapping between infinite-dimensional function spaces. With basic to deep neural network architectures being data-driven, i.e. dependent on quality data for prediction, neural operators provide robustness by offering generalization among different resolutions despite low quality data. Graph neural operators are a variant that utilize graph networks to learn this mapping between function spaces. The aim of this research is to identify robust machine learning architectures that integrate spatial information to predict tissue types. Under this notion, we propose a study incorporating various graph neural network approaches to validate the efficacy of applying neural operators towards prediction of brain regions in mouse brain tissue samples as a proof of concept towards our purpose. We were able to achieve an F1 score of nearly 72% for the graph neural operator approach which outperformed all baseline and other graph network approaches.

cs.LG↗

arXiv2021

Yield strength insensitivity in a dual-phase high entropy alloy after prolonged high temperature annealing

Recent studies of FeMnCoCr-based high entropy alloys have demonstrated uncommon deformation behaviors such as transformation-induced plasticity, which were largely believed to be restricted to select families of steels. Coupled with the potential for entropy stabilization of high symmetry phases at high temperatures, this system represents a promising class of materials for structural applications in extreme environments. Yet, transformation-induced plasticity mechanisms are notably sensitive to microstructure parameters and the literature offers examples of deleterious decomposition of high entropy alloys under heat treatment, which raises concerns of resiliency in mechanical performance. Here, we evaluate the evolution of microstructure and mechanical properties of a FeMnCoCr high entropy alloy after prolonged heat treatment at high temperature. Microstructures are found to retain their characteristic austenite/martensite features, with parent face-centered cubic grains partitioned by hexagonal close-packed laths after heat treatment at 1200 C for up to 48 hours. Results of mechanical testing reveal an unusual insensitivity of this alloy to grain growth-induced weakening effects. Namely, the yield strengths of FeMnCoCr samples are observed to remain constant across all heat treatment conditions, despite a near four-fold increase in the grain size. Close examination of post-heat treatment microstructures reveals a dramatic decrease in the inter-lath spacing at longer durations, which segments parent austenite grains. This crystal partitioning counteracts conventional grain growth-induced weakening by introducing additional barriers for dislocation pile-up. These results offer new insights into the mechanical resiliency of this transformation-induced plasticity high entropy alloy under prolonged high temperature heat treatment.

cond-mat.mtrl-sci↗

arXiv2020

Approximate Nearest Neighbor Negative Contrastive Learning for Dense Text Retrieval

Conducting text retrieval in a dense learned representation space has many intriguing advantages over sparse retrieval. Yet the effectiveness of dense retrieval (DR) often requires combination with sparse retrieval. In this paper, we identify that the main bottleneck is in the training mechanisms, where the negative instances used in training are not representative of the irrelevant documents in testing. This paper presents Approximate nearest neighbor Negative Contrastive Estimation (ANCE), a training mechanism that constructs negatives from an Approximate Nearest Neighbor (ANN) index of the corpus, which is parallelly updated with the learning process to select more realistic negative training instances. This fundamentally resolves the discrepancy between the data distribution used in the training and testing of DR. In our experiments, ANCE boosts the BERT-Siamese DR model to outperform all competitive dense and sparse retrieval baselines. It nearly matches the accuracy of sparse-retrieval-and-BERT-reranking using dot-product in the ANCE-learned representation space and provides almost 100x speed-up.

cs.IR↗

arXiv2019

Stochastic Geometry Modeling of Cellular V2X Communication Over Shared Channels

To overcome the limitations of DSRC with short range, non-supportability of high density networks, unreliable broadcast services, signal congestion and connectivity disruptions, cellular vehicle-to-everything (C-V2X) communication networks, standardized in 3rd Generation Partnership Project (3GPP) Release 14, have been recently introduced to cover broader vehicular communication scenarios including vehicle-to-vehicle (V2V), vehicle-to-pedestrian (V2P) and vehicle-to-infrastructure/network (V2I/N). In C-V2X, vehicles can directly communicate over PC5 based dedicated sidelinks called direct mode or V2V communication. However, high vehicle densities may require reuse of cellular spectrum for V2V. Moreover, infrastructure mode communication through V2I/N links can augment V2V communication by enhancing communication range and reliability for enhanced safety along with consistent performance under traffic congestions. Motivated by the stringent connection reliability, spectral efficiency, and coverage requirements in C-V2X, this paper presents the first comprehensive and tractable analytical framework for performance of C-V2X networks over shared V2V and cellular uplink channels, where the transmitting vehicles can deliver their information via infrastructure or direct mode, based on their distances, propagation environments and the bias factor. By practically modeling the vehicles on the roads using the doubly stochastic Cox process and the base-stations, we derive new association probabilities, new success probabilities of infrastructure and direct mode, and overall success probability of the C-V2X communication over shared channels, which are validated by the simulations results. Our results reveal the benefits of our proposed model (possibility of selecting both direct and infrastructure modes over shared channels) compared to V2V network in terms of success probability.

cs.NI↗

arXiv2018

On the Secrecy Performance of SWIPT Receiver Architectures with Multiple Eavesdroppers

Physical layer security (PLS) has been shown to hold promise as a new paradigm for securing wireless links. In contrast with the conventional cryptographic techniques, PLS methods exploit the random fading in wireless channels to provide link security. As the channel dynamics prevent a constant rate of secure communications between the legitimate terminals, the outage probability of the achievable secrecy rate is used as a measure of the secrecy performance. This work investigates the secrecy outage probability of a simultaneous wireless information and power transfer (SWIPT) system, which operates in the presence of multiple eavesdroppers that also have the energy harvesting capability. The loss in secrecy performance due to eavesdropper collusion, i.e., information sharing between the eavesdroppers to decode the secret message, is also analyzed. We derive closed-form expressions for the secrecy outage probability for Nakagami-$m$ fading on the links and imperfect channel estimation at the receivers. Our analysis considers different combinations of the separated and the integrated SWIPT receiver architectures at the receivers. Numerical results are provided to validate our analysis.

cs.IT↗