Pinned Repositories
194-Face-Point-Detection-CNN
This Model will detect194 points on the face and has scored a loss = 0.05 and val_loss=50 on training and testing images.
2-Dimensional-Maze-Route-Q-Learning
This project is using traditional Q Learning to solve a game without Deep Learning
68-Face-point-level-1
Detect 68 points on face live (i-BUG). This is running two neural networks which are very light as it runs on 30% of Intel i7-7700k CPU. Please run liveDetection.py for live detection on face. Model score= loss:5 val_los:6 on 100X100 grayscale image.
Appliances-energy-prediction-data
Data sets and scripts for the publication in Energy and Buildings
Bitcoin-Price-Prediction-RNN
This Application will predict Bitcoin price for the next 1m, 15m, 30m and 1hour. This Application uses Crypto Compare API (All Rights Reserved). The Application is Built with the help of Tensorflow and Keras powered by NVIDIA GTX 1060 GPUs
Cartpole-DQN
This is a solution for the OPEN AI gym project cart pole using Deep Q Learning
COVID-19-Simulation-Analysis-1.github.io
COVID-19 Simulation and Analysis on Unrestricted Environment. This simulation only has the original varaint of COVID-19 and focusing on population density and mask policies around the world.
COVID-19-Variants-Prevention-Simulation-Analysis-2.github.io
COVID-19 Simulation and Analysis on Unrestricted Environment. This simulation only has the simulation on the original variant and the UK variant and analyzing the difference in spread and preventions.
covid19-Analysis-City.github.io
Basic Analysis of Covid-19 spread in Sydney.
Google-Natural-Language-Dataset-Question-Answering-Personal-Research-Project
The goal for this project is to answer questions from a large amount of text which was not possible before. Have you ever wanted to answer questions from a book without reading it. Using these models it will make answering question of book by simply scanning through the large text document and answering questions.
NiksanJP's Repositories
NiksanJP/Bitcoin-Price-Prediction-RNN
This Application will predict Bitcoin price for the next 1m, 15m, 30m and 1hour. This Application uses Crypto Compare API (All Rights Reserved). The Application is Built with the help of Tensorflow and Keras powered by NVIDIA GTX 1060 GPUs
NiksanJP/194-Face-Point-Detection-CNN
This Model will detect194 points on the face and has scored a loss = 0.05 and val_loss=50 on training and testing images.
NiksanJP/68-Face-point-level-1
Detect 68 points on face live (i-BUG). This is running two neural networks which are very light as it runs on 30% of Intel i7-7700k CPU. Please run liveDetection.py for live detection on face. Model score= loss:5 val_los:6 on 100X100 grayscale image.
NiksanJP/Google-Natural-Language-Dataset-Question-Answering-Personal-Research-Project
The goal for this project is to answer questions from a large amount of text which was not possible before. Have you ever wanted to answer questions from a book without reading it. Using these models it will make answering question of book by simply scanning through the large text document and answering questions.
NiksanJP/2-Dimensional-Maze-Route-Q-Learning
This project is using traditional Q Learning to solve a game without Deep Learning
NiksanJP/Appliances-energy-prediction-data
Data sets and scripts for the publication in Energy and Buildings
NiksanJP/Cartpole-DQN
This is a solution for the OPEN AI gym project cart pole using Deep Q Learning
NiksanJP/COVID-19-Simulation-Analysis-1.github.io
COVID-19 Simulation and Analysis on Unrestricted Environment. This simulation only has the original varaint of COVID-19 and focusing on population density and mask policies around the world.
NiksanJP/COVID-19-Variants-Prevention-Simulation-Analysis-2.github.io
COVID-19 Simulation and Analysis on Unrestricted Environment. This simulation only has the simulation on the original variant and the UK variant and analyzing the difference in spread and preventions.
NiksanJP/covid19-Analysis-City.github.io
Basic Analysis of Covid-19 spread in Sydney.
NiksanJP/data-science-wg
SF Brigade's Data Science Working Group.
NiksanJP/Gym-Cartpole-Q-Learning
This is a simple Q Learning example with open AI gym. Therefore, it's efficiency is not enough to meet the requirements. Unlike same Q Learning the Deep Q Learning uses Neural Networks and will work more efficiency
NiksanJP/ImageAI-Live-Detection
Simple detection technique using ImageAI. This uses darknet's YOLOv3 which is only compatible with Linux and MAC. I decided to use Opencv and open each frame and process it with Image AI image detection.
NiksanJP/Leetcode
Solutions to LeetCode problems; updated daily. Subscribe to my YouTube channel for more.
NiksanJP/Leetcode-Questions-And-Answers
NiksanJP/models
Models and examples built with TensorFlow
NiksanJP/Number-Recognition-via-drawing
This program is a Machine learning program that uses Tensorflow with Keras models. This ML program is trained using the MNIST dataset. Unfortunately the MNIST dataset does not give real life images and the trained model is not good enough for real life images.
NiksanJP/One-Dimensional-Q-Learning-and-Reinforcement-Learning
Bellman equation gives a reward for every correct step. The Reinforcement Learning (RL) program I wrote follows this principle of the equation. The board is not small so it might take some to learning.
NiksanJP/Puzzel-and-dragon-game
This is an old project, I made this project when I jut started programming. There might be some errors and stupid decisions made when making this code. Feel free to use this.
NiksanJP/Snake-Deep-Q-Learning
Uses Deep Q Learning to find the best path to reward. The snake retrains itself every time and if a wrong move is made, the snake fits the state with a correct move.
NiksanJP/Song-Popularity-Analysis-and-predictions
Can we predict if the song is going to be a hit. Lets analyse and predict using Spotify's Web API.