Custom Loss Functions and Evaluation Metrics for XGBoost and LightGBM

Detect keypoints at a football pitch

Used the Functional API to built custom layers and non-sequential model types in TensorFlow, performed object detection, image segmentation, and interpretation of convolutions. Used generative deep learning including Auto Encoding, VAEs, and GANs to create new content.

Utilities for easy use of custom losses in CatBoost, LightGBM, XGBoost.

Temporal Attention Fusion Network with Custom Loss Function for EEG-fNIRS Classification

An HR predictive analytics tool for forecasting the likely range of a worker’s future job performance using multiple ANNs with custom loss functions.

Neural Style Transfer implementation for images and videos using Tensorflow2.

This repository contains code used for the numerical experiments in the Supervised Learning for Integrated Forecasting and Inventory Control paper by Joost F. van der Haar, Arnoud P. Wellens, Robert N. Boute and Rob J.I. Basten.

custom loss functions

Simulation of autonomous driving by deep learning with GPU accelerated computing on Google Cloud. Augmentation and custom loss-functions allow self-driving vehicles to maintain lanes, for single trips around virtual tracks, in speed-controlled environments.

Deep-Learning approach for generating Fair and Accurate Input Representation for crime rate estimation in continuous protected attributes and continuous targets.

Predict crypto prices using neural nets + LLM-powered custom loss functions on real Chainlink data (ETH/USD)

This repository is a testament to my growth and understanding of deep learning. It highlights foundational skills in TensorFlow, demonstrates hands-on applications like car price prediction, and culminates in an advanced medical imaging project for malaria diagnosis.

Multi-task deep learning for predicting house price and category using PyTorch and Lightning

Building Advanced Neural Networks with Tensorflow: A Deepdive

Facial Landmark Detection Using Knowledge Distillation-Based Neural Networks

This code is a custom implementation of the Supervised Contrastive Learning paper (https://arxiv.org/abs/2004.11362).