Aakash Tripathi's Projects
Profile Repository
This is our Flask-based, Microservices Architecture platform that brings automation to data preprocessing and machine learning, simplifying your data science journey.
This repository contains a BigInteger library implemented in C++, which provides support for various arithmetic operations and functions for working with large integers.
This repository contains a Jupyter Notebook that explores various clustering techniques applied to the Fashion MNIST dataset like K-Means, Hierarchical,etc.
This repository contains a custom implementation of a generic Deque (double-ended queue) data structure. A deque is a versatile data structure that allows efficient insertion and deletion operations at both ends of the queue.
This repository contains an implementation of the Decision Tree algorithm from scratch using various impurity methods such as Gini index, entropy, misclassification error, etc.
This repository contains the digit recogniser datasets and the analysis in a .ipnb notebook as well as the submission.csv file. The Predictions are done using Convolutional Neural Networks. The overall score for the predictions of this model is 0.98400 in the actual compititions. Imporovement in the model which improves the accuracy are always welcome!!!
Ever tried writing/drawing in air? Running this python script based on opencv will track any green/purple object of a specific hsv and will track that object.
This GitHub repository explores the importance of MLP components using the MNIST dataset. Techniques like Dropout, Batch Normalization, and optimization algorithms are experimented with to improve MLP performance. Gain a deeper understanding of MLP components and learn to fine-tune for optimal classification performance on MNIST.
This repository contains a Jupyter Notebook that implements Gaussian Mixture Model (GMM) for semantic segmentation and background extraction. GMM class is implemented from scratch without using any libraries like sklearn.
This repository contains the implementation of Gaussian Naive Bayes from scratch in a Jupyter Notebook. Gaussian Naive Bayes is a simple and effective algorithm for classification tasks. It is based on Bayes' theorem with the assumption of independence between the features.
This game, Kaoaa, has been developed using pure HTML, CSS, and JavaScript for the purpose of learning pure JavaScript handling techniques. The primary goal of this project is to provide a valuable learning experience in handling JavaScript.
This Jupyter Notebook demonstrates the implementation of a K-Nearest Neighbors (KNN) algorithm using the concept of nearest neighbors without using direct classifiers. It also includes exploratory data analysis (EDA) and comparison of three classifiers.
This repository implements N-gram language modeling with Kneser-Kney and Witten Bell smoothing techniques, including an in-house tokenizer. It also features a neural model with LSTM architecture and calculates perplexities for comparing language and neural models.
This repository contains a Jupyter notebook that implements Linear Regression using Gradient Descent from scratch. The notebook also includes a comparison of the results with the scikit-learn implementations of Linear, Lasso, and Ridge Regression by plotting graphs.
This GitHub repository provides an implementation of the paper "MAGNET: Multi-Label Text Classification using Attention-based Graph Neural Network" . MAGNET is a state-of-the-art approach for multi-label text classification, leveraging the power of graph neural networks (GNNs) and attention mechanisms.
This project is a peer-to-peer file sharing network that enables users to share, download, and remove files from the group they belong to. Download happens parallelly with multiple pieces from multiple peers. It follows a similar concept to popular file sharing protocols like BitTorrent found on the internet.
This repository contains a Jupyter notebook implementing the Multinomial Naive Bayes algorithm from scratch for an email classification task of SPAM or HAM. The notebook also includes a comparison of the results obtained with the scikit-learn implementation of Multinomial Naive Bayes.
This repository contains an implementation of a neural sequence model (RNN, LSTM, GRU, etc.) to tokenize and tag sentences with the correct part-of-speech (POS) tags.
A Simple Notes App which can get a string , put it in mongoDB database and then fetches it to show to a webpage using Node.js and Express.js used for routing. This project is for looking at basic mongoDB connections with our Web Application.
This repository contains a jupyter notebook and a csv file which contains batting records of all ODI players ever played ODI cricket. This task was accomplished through Web scraping in python using libraries like BeautifulSoup , Pandas and numpy.
This repository contains a Jupyter Notebook that implements PCA (Principal Component Analysis) from scratch for facial recognition. It demonstrates the steps involved in PCA, including eigenface computation and accuracy comparisons for different components.
This project is based on the datasets provided by a kaggle competition. The provided datasets includes: 1)train.csv 2)test.csv 3)sampleSubmission.csv -Using Different Classification techniques like naive_baeyes,RandomForest,etc , i have predicted the categories of crimes on the test.csv dataset.
This is the analysis and predictions of a traffic dataset provided by IIT Madras Alumini assosiation in a competition called Sangam Hackathon 2019 conducted in the month of july-August 2019. This consists of a Time Series dataset on which we apply regression problems