Search
6 articles for “TRACE benchmark”
-
TRACE: Tracking the Loss of Memory Provenance in LLM Agents
Abstract: AI agents that carry memory across sessions gain better personalization and decision-making, but this persistence opens a serious security gap. When an agent repeatedly condenses earlier interactions into compact “lessons,” the trail linking each lesson back to the interaction that produced it gradually fades. We term this effect Reflective Attribution Collapse (RAC): the progressive loss of provenance and forensic traceability that results from repeated memory reflection. Under RAC, a malicious …
Published in Journal of Advancements in Robotics · Vol. 13, Issue 2, 2026 · pp. 33–49 Read article
-
Neuro-Symbolic Agentic AI for Autonomous Scientific Discovery: Integrating Deep Reinforcement Learning, Quantum Simulation, and XAI-Audited LLM Hypothesis Generation in Drug Target Identification
Abstract: The exponential growth of multi-omics data and the increasing complexity of disease-associated protein interactomes have rendered conventional drug target identification pipelines computationally and epistemologically inadequate. This paper presents the Neuro-Symbolic Agentic AI for Scientific Discovery (NS-AASD) framework, a unified architecture that cohesively integrates deep reinforcement learning (DRL) exploration strategies, variational quantum simulation (VQS) of protein conformational dynamics, and XAI-audited large language model (LLM) hypothesis generation within an autonomous scientific discovery …
Published in Research and Reviews : Journal of Computational Biology · Vol. 15, Issue 2, 2026 · pp. 15–24 Read article
-
Impact of Groundwater Pollution on Carcinogenic and Non-Carcinogenic Health Effects: A Review
Abstract: Groundwater is a critical source of drinking water in many regions worldwide, particularly in South and Southeast Asia, where dependence on untreated groundwater remains high. However, increasing groundwater contamination poses serious risks to human health. This review synthesizes recent peer-reviewed studies to examine the sources, distribution, and health impacts of major groundwater contaminants, with a focus on both carcinogenic and non-carcinogenic risks. Emphasis is placed on fluoride, nitrate, chromium, and …
Published in Research & Reviews : Journal of Ecology · Vol. 15, Issue 1, 2026 · pp. 40–46 Read article
-
Reward-Aligned Reinforcement Learning from Human Feedback for Emotion-Sensitive Large Language Model Therapists: Balancing Empathetic Engagement, Boundary Safety, and Clinical Accountability
Abstract: The deployment of large language models (LLMs) in mental health therapy contexts introduces a critical alignment challenge: these systems must simultaneously cultivate genuine empathic rapport, observe clinically grounded safety boundaries, and remain auditable under institutional accountability frameworks. Existing reinforcement learning from human feedback (RLHF) pipelines optimize for a scalar reward signal that is demonstrably insufficient for the multi-objective, temporally extended nature of therapeutic conversation. This paper presents RA-RLHF-T (Reward-Aligned RLHF …
Published in International Journal of Tropical Medicines · Vol. 3, Issue 2, 2026 Read article
-
Genomic Selection for Grain Yield in Wheat Using Machine Learning on DArT Molecular Markers: A Comparative Evaluation Across Multi-Environment Trials
Abstract: Genomic selection (GS) predicts complex quantitative traits directly from genome-wide molecular markers, bypassing the need for extensive phenotypic trials and accelerating plant breeding cycles. We conducted a comparative evaluation of seven regression approaches — ridge regression (the machine-learning equivalent of RR-BLUP), Lasso, Elastic Net, Partial Least Squares, linear Support Vector Regression, Random Forest, and Gradient Boosting — for predicting grain yield from 1,279 Diversity Array Technology (DArT) molecular markers genotyped …
Published in Research and Reviews : Journal of Computational Biology · Vol. 15, Issue 2, 2026 Read article
-
Blockchain Enabled IoT System for Tamper Proof Monitoring of Polymer Composite Manufacturing Quality
Abstract: Ensuring real-time process compliance in resin-based polymer composite manufacturing remains a persistent challenge due to non-linear material behaviors, unpredictable curing dynamics, and fragmented sensor data pipelines. Traditional centralized monitoring architectures struggle to guarantee data integrity, auditability, and adaptive response under high-frequency environmental fluctuations. Most existing frameworks fall short in unifying trust, traceability, and time-critical decision-making particularly during critical cure-phase deviations due to limited integration of blockchain with intelligent sensor systems. …
Published in Journal of Polymer & Composites · Vol. 13, Issue 6, 2025 · pp. 126–145 Read article