Publications · 152+
Peer-reviewed work from SPARKS Lab researchers across the leading venues in machine learning, vision, language, and applied AI.
Structured knowledge graphs constrain LLM hallucinations through dynamic retrieval and iterative self-correction across multi-hop chains.
Abdullah A., Sharma R., Krishnan V.
A clinically validated framework combining EHR, imaging and waveform data with attribution for clinician trust.
Sharma R., Abdullah A., Iyer M.
Decentralized framework for adaptive vision-model updates across constrained IoT hardware in real time.
Iyer M., Abdullah A., Patel S.
Open release of a 7B-parameter foundation model with strong cross-lingual transfer across low-resource Indic languages.
Reddy V., Abdullah A., et al.
40% latency reduction on transformer inference with no accuracy degradation on a benchmark suite of 12 tasks.
Patel S., Abdullah A.
Empirical study of representation collapse in deep knowledge embeddings and methods to prevent it.
Abdullah A., Menon A.
Methodology for adapting large multilingual models to low-resource South Asian languages efficiently.
Krishnan P., Joshi N.
Temporal transformer architecture deployed at three Indian tertiary care hospitals with 94% precision.
Sharma R., et al.
Diagnostic toolkit for probing what vision-language models truly know vs. memorize.
Menon A., Iyer M.