Skip to content
View Simhaatt's full-sized avatar
0 followers · 1 following

Highlights

  • Pro

Block or report Simhaatt

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Simhaatt/README.md

Simhaa T. T.

Research student, Department of Electrical & Electronics Engineering, SSN College of Engineering, Chennai. I work on machine learning for physical and remote-sensing data, with a bias toward methods that are reproducible by a stranger: frozen configs, deposited artifacts, and a one-command check that the numbers in the paper are the numbers the code produces.

  • Interests: hyperspectral image analysis · graph-based / training-free methods · time-series forecasting for battery degradation · multi-objective optimisation
  • Currently: preparing manuscripts on spectral-spatial region discovery and matched-budget multi-objective forecasting
  • Reach me: ORCID 0009-0006-9041-1103

Research code

Project What it is Artifact
Graph-RMS Training-free, class-count-free spectral–spatial region discovery in hyperspectral images. Sparse reciprocal graph → damped diffusion → robust mean-shift modes → reference-free consolidation. Nine scenes, Trento held out. DOI
battery-moo-recursive-forecasting Recursive LSTM forecasting of capacity and resistance under a matched-budget NSGA-II / NSGA-III / random-search comparison. Headline: 92.5% fewer parameters at statistically indistinguishable accuracy (12,600-trial log included). DOI

Both repositories are built to be re-run by a stranger: the reported tables and figures regenerate from deposited artifacts in minutes, on CPU, with no dataset download and no model retraining (python analysis/reproduce_all.py in the battery repository; scripts/export_tables.py and scripts/generate_figures.py in Graph-RMS).

Applied / systems work

Project What it is
money_muling_detection Graph-based fraud detection over transaction networks — PageRank, betweenness, cycle detection, Louvain communities and temporal smurfing windows behind a FastAPI + React app. Live demo.
ocr Document OCR, field mapping and fuzzy verification (PaddleOCR + FastAPI) wired to a MOSIP pre-registration flow.

Toolbox

Python · PyTorch · NumPy / SciPy · scikit-learn · NetworkX · FastAPI · NSGA-II / NSGA-III · TypeScript / React · pytest · LaTeX

Pinned Loading

  1. ocr ocr Public

    Document OCR, field mapping and fuzzy verification (PaddleOCR + FastAPI) wired to a MOSIP pre-registration flow

    Python

  2. battery-moo-recursive-forecasting battery-moo-recursive-forecasting Public

    Recursive LSTM forecasting of battery degradation under matched-budget NSGA-II / NSGA-III search — 92.5% fewer parameters at indistinguishable accuracy (DOI 10.5281/zenodo.22067837)

    Jupyter Notebook

  3. Graph-RMS Graph-RMS Public

    Training-free, class-count-free spectral-spatial region discovery in hyperspectral images — reproducibility package (DOI 10.5281/zenodo.21876447)

    Python

  4. money_muling_detection money_muling_detection Public

    Graph-based money-mule detection over transaction networks — PageRank, cycle detection, Louvain communities and temporal smurfing behind a FastAPI + React app

    TypeScript 1