Publications and preprints. Full list on Google Scholar.
V. Pandey, R. Tewari, R. H. Laskar · arXiv preprint · arXiv:2606.21892
Cross-modal transformer fusion combining soil imagery, tabular nutrient profiles, and geospatial raster inputs for interpretable multi-task crop recommendation under real-world telemetry.
V. Pandey, G. Singh · arXiv preprint · arXiv:2606.09287
A geometric analysis of how token representations move through transformer depth, characterizing curvature, drift, and separability of trajectories across layers.
V. Pandey, R. H. Laskar, R. Tewari · arXiv preprint · arXiv:2605.13464
A three-stage pipeline for diabetes detection, subtype discrimination, and cognitive-metabolic hypothesis testing with interpretable, SHAP-grounded per-patient attribution.
V. Pandey, G. Singh · arXiv preprint · arXiv:2605.11196
A variational recurrent memory update with provably unit Jacobian spectral norm, enabling stable long-context reasoning without softmax's quadratic cost.
V. Pandey, R. Das, D. Biswas · arXiv preprint · arXiv:2509.01344 · Cited by 7
An integrated deep learning system that combines soil image analysis with nutrient profiling to recommend crops under noisy, field-collected inputs.
V. Pandey, D. Biswas · arXiv
A spiking formulation of decision transformers using LIF neurons, surrogate-gradient training, phase-coding, and dendritic routing for low-power sequence control.