Sparse Intelligent Systems
Efficient representations of complexity using sparsity, hierarchy, and validation-driven reduction. Designed to scale with practical compute.
Learn more →Building efficient intelligent systems from algorithms, hardware, and real-world data.
I work on scalable AI architectures, large-scale spatial clustering, scientific instrumentation, embedded intelligence, bioinformatics data mining, and planetary exploration tools—spanning decades of research and real deployments.
Research themes
Curated view for fast understanding. Detailed material is in the Archive.
Neural Architectures
Stochastic neuron concepts, partially connected feedback networks (associative memory), and hierarchical “net-of-nets” architectures.
Learn more →Scientific Instrumentation
Intelligent instruments combining electronics, sensing, and AI—3D imaging microscopy, radiation signal discrimination, and embedded payload modules.
Learn more →Bioinformatics & Data Mining
Protein secondary structure prediction, biological database utilities, and proteomic data mining using hybrid knowledge + neural methods.
Learn more →Planetary Exploration Tools
Crater detection, terrain browsing and visualization, astro-material analysis, and tools supporting planetary science workflows.
Learn more →Embedded & Distributed AI
Sensor networks, low-power analytics, and distributed computation concepts to push intelligence closer to real-world signals.
Learn more →Featured projects
Six highlights. Each has a dedicated project page with figures, results, and downloads.
Scalable Algorithms
Large-Scale Spatial Clustering
Near-linear clustering on very large spatial datasets; designed for robust behavior and practical runtime on standard computing platforms.
Open →Hierarchical Learning
Net-of-Neural-Networks Architecture
Modular architecture where pre-trained networks act as higher-order computational units inside larger adaptive systems.
Open →Foundational Models
Stochastic Neuron & Hardware-Friendly Learning
Probabilistic neuron behavior inspired by biological firing, along with learning rules and VLSI-style implementation thinking.
Open →Imaging
Reflective Light Microscope for 3D Reconstruction
Low-cost 3D imaging using focus stacking and intelligent focus/defocus detection for thick and irregular surfaces.
Open →Planetary Science
Crater Simulation & Terrain Browsers
Tooling for planetary surface analysis, visualization, and crater-related workflows for research and teaching labs.
Open →Signal Processing
Radiation Counting Efficiency Improvement
Noise signature discrimination and anti-coincidence ideas to enhance weak-signal detection in noisy environments.
Open →Selected publications
A short list on the homepage. Full list with PDFs/DOIs on the Publications page.
Knowledge base and neural network approach for protein secondary structure prediction
A Neural Network Tool Box using C++
Rainfall prediction using neural network based approaches