Radio Frequency Machine Learning with PyTorch

Motor Imagery EEG Signal Classification Using Random Subspace Ensemble Network

Code for "Complex-Valued Depthwise Separable Convolutional Neural Network for Automatic Modulation Classification"

Detection of human emotions from eeg signals using the amigos dataset

This repository contains the implementation of the modified convolutional neural network (MCNN) for classification of Frequency-Hopping Spread Spectrum Signals using the time-frequency representations (TFRs) as an images.

Production-grade ensemble framework combining XGBoost, PyTorch & Sklearn - 70%+ test coverage with Optuna optimization for time-series prediction

AtrialFibrillation-Classification-SMoLK is a deep learning project designed to classify arrhythmias, specifically focusing on atrial fibrillation (Afib), using the MIT-BIH Arrhythmia Database. The project leverages a neural network model with learned filters to achieve high classification accuracy across multiple classes.

A transform with at most 25 code lines proved so far a convenient substitute for more sophisticated transforms (e.g. MFCCs) in various signal recognition problems. Particularly useful in TinyML context

Classification of biosignals, including EEG and MEG, through feature extraction in the time and frequency domains using machine learning methods.

📈 Análise e processamento de sinais com SVD

Real-time robot collision detection using supervised learning methods

VBS Power Line Fault Detection (Kaggle challenge)

"Machine learning project for classifying music into genres using audio feature extraction and classification models."

Deep Leaning for Classification of drone signals.

A deep learning project classifying ECG signals using time series models

One of the firsts dataset level explanability libraries for 1d signal using GRAD-CAM++

An automated GTM intelligence platform for aggregating, categorizing, and generating insights from Revolut-related signals (news, funding, partnerships, etc.). The platform scrapes data from Google News, classifies it into GTM themes, and delivers concise summaries and actionable recommendations.

This project uses time-domain EMG signal features to classify gender (male/female) via various machine learning models. It explores multiple resampling techniques (SMOTE, ADASYN, etc.) to handle class imbalance and compares model performance using cross-validation and test accuracy. Visual plots a