Shayan Ali Akbar's repositories
CNN_Classification
The repository implements the a simple Convolutional Neural Network (CNN) from scratch for image classification. I experimented with it on MNIST digits and COIL object dataset.
3D_Tree_Reconstruction
The repository implements methods that deal with constructing accurate 3D reconstructions of dormant apple fruit trees. The depth images are gathered from real orchards and point clouds are generated. The point clouds are then merged together using Iterative Closest Point ICP algorithm. Then shape (circle, semi circle, cylinder) fitting is performed to get 3D model of the trees.
Clustering_DimReduction
The implementations in this repository deal with clustering and dimensionality reduction for MNIST digits dataset. Kmeans clustering algorithm is implemented. Also different hierarchical clustering algorithms are tested. We also play with the PCA and TSNE embeddings of the MNIST dataset.
sakbarpu.github.io
Welcome to my github main page.
TraditionalClassifiers
This repository implements the basic machine learning classifiers for the problem of Yelp reviews classification. We assume the problem to be a binary classification problem. The models implemented are Naive Bayes, Logistic Regression, Support Vector Machine (linear), Decision Trees, Bagged Decision Trees, Random Fforests, and Boosted Decision Trees.
bme_project
This is a multiprocessing implementation of word2vec algorithm in Python from scratch without using any machine learning libraries.
BugLocalization
The repository implements code for information retrieval based bug localization. The open source Terrier search engine is extended to implement the novel algorithms that deal with performing IR based bug localization.