NikolasMarkou/dl_techniques
Advanced deep learning learning techniques, layers, activations loss functions, all in keras / tensorflow
Lifelong technologist, pragmatic software engineer and experienced machine learning practitioner.
Advanced deep learning learning techniques, layers, activations loss functions, all in keras / tensorflow
A Finite State Machine hybrid with Large Language Models
Rebuild a repository's documentation map by generating self-contained README.md and CLAUDE.md
Building a multiscale variational autoencoder
A Claude Code skill state-machine driven iterative planning and execution protocol for complex coding tasks
Exploring a new initialization on deep learning based on an isosceles simplex scheme
Collection and notes on AI standards
Claude Code for python coding
Claude code skill and notes for auditing ML solutions
Claude skill for reverse engineering systems
Claude Skill for building agents
bio
A production-grade Python package for multi-account cash flow forecasting with layered architecture, transfer netting, cash flow decomposition, and LLM-ready explainability output.
The Leek group guide to data sharing
Zerocoin: C++ lib for anonymous extension to bitcoin
A small opencv based utility that is used to extract images from pictures or videos
This is parth of the Coursera Data Science Specialization course
Caffe: a fast open framework for deep learning.
Fast R-CNN
Machine learning, computer vision, statistics and general scientific computing for .NET
Bardcode reader
Demonstration of the S.O.L.I.D principles in C++
Useful bash snippets collected over time
SqueezeNet: AlexNet-level accuracy with 50x fewer parameters
Deep learning pattern recognition on the famous mnist dataset
A modern C++11 header only implementation of a generic n-dimensional histogram with no dependencies
Exploring the predictive power of the inverse of ReLU cutouts
ONNX-TensorRT: TensorRT backend for ONNX
A extendible and robust way of deploying production level deep learning models
Protocol Buffers - Google's data interchange format