§·IIIThe Mind
Research
Research in pursuit of intelligence that deserves trust: multi-agent systems that keep people at the centre of the factory floor, retrieval that refuses to erase an exception, and models judged by what they cost in watts as well as by what they get right.
1
paper out
1
in preparation or review
The School of Athens
Raphael
Apostolic Palace, Vatican, 1509–11
The library
2 folios- PublishedConference · 2026
SMAHI-ER: Symbiotic Multi-Agent Hybrid Intelligence for Resilient Human-Machine Manufacturing
Febin K Renu
IEEE IEHNS 2026
SMAHI-ER is a human-centric Industry 5.0 framework that models the manufacturing ecosystem as a dynamic hypergraph, joining Human Digital Twins, Machine Cognitive Twins and Ethical Meta-Agents through relationships of trust, communication, dependency and ethical alignment. Graph neural networks give the system emergent cognition; generative diffusion reinforcement learning synthesises adaptive strategies; causal SHAP keeps its ethical decisions interpretable; and federated learning lets factories collaborate without surrendering their data. Evaluated across more than 50 disruption scenarios with 30 Monte Carlo runs each, it sustains 88.1% overall equipment effectiveness, a 1.2% defect rate and 91.5% control of worker stress, with 2.15× the emergent intelligence of baseline approaches.
- Industry 5.0
- multi-agent systems
- hypergraph GNN
- diffusion RL
- causal SHAP
- federated learning
- digital twins
- In preparationPreprint · 2026
Beyond Source Selection: Detecting and Resolving Conditional Knowledge Conflicts in Retrieval-Augmented Generation
Febin K Renu
Final-year research, Karunya Institute of Technology and Sciences
Conflict-aware retrieval-augmented generation resolves disagreements between passages by selecting the most credible source. For a large and common class of disagreement this is the wrong operation: a general rule and its exception are both true under different scopes, and selection silently deletes the exception together with the condition that governs it. The work treats conditional conflicts as a first-class type, extracts the governing conditions, and composes a scoped answer that preserves every valid branch. It introduces the Suppression Rate metric suite, which makes the deletion measurable where answer-correctness cannot, and a benchmark of naturally occurring rule-and-exception pairs on which to measure it.
- retrieval-augmented generation
- knowledge conflicts
- conditional reasoning
- NLI
- evaluation