Deep Learning

Neural Machine Translation(NMT)

encoder-decoder architecture was implemented with different recurrent units and layers to translate Engligh to Persian and English to Spanish

Generative Adversarial Network (GAN)

Convolutional GAN was implemented using tensorflow to generate simpson faces

Stock Market Prediction

Various recuurrent units was compared to each other in time-series prediction task.

Convolutional Neural Network(CNN)

CNN hyper-parameters e.g., kernel size, convolutoin layers, pooling meethods, dropout rate were experimented on STL dataset.

Self-Organizing Maps (SOM)

SOM and growing SOM(GSOM) was implemented from scracth for clustering. different neighbourhood topologies were compared to each other. also SOM was used to reduce dimensionality of high-dim data

RBF and EBF Neural networks

Radial Basis Function and Eliptical Basis Function networks were used for classification and regression

Deep Feed-Forward Neural Network

DNN was used to classify MNIST dataset. various hyper-parameters were experimented. implemented using tensorflow's estimator API.

Information Retrieval

Information Retrieval on Hamshahri corpus

in this project, vector space model(TF-IDF), unigram language model, translation model and word2vec was implemented.

Recommender System

in this project collaborative filtering and matrix factorization methods were implemented for recommender systems on Amazon dataset

Topic Modeling

In this project latent dirichlet allocation(LDA) was used for topic modeling. Also topic tag correlations were calculated using Bayes rule and cannonical correlation(CCA) on Wiki10-31K and EUR-Lex. performace was compared with neural models.

Image Proccessing

Basic Image manipulation

contrast stretching, histogram equalization, histogram matching, template matching, and filters were practiced

Fourier transform and frequency domain

some image manipulation was performed in frequency domain such as denoising, template matching, iamge reconstruction.

Compression and morphology

huffman coding, arithmetic coding, LZW, differential coding, predictive coding and morphological operators were practiced

Statistical Machine Learning

statistics and probabilities fundamentals

In this course statistical and mathematical side of machine learning e. g., probability theory, distribution estimation, hypothesis testing were practiced

Machine Learning

Machine learning algorithms implementation

In this course different traditional machine learning algorithms such as gradient descent, K Nearest Neighbor(KNN), SVM, DBSCAN, hierarchical clustering were implemented