SHAHROKH SHAHI | projects

This section is still under construction as I am currently working on migrating everything from my older personal website to this online portfolio. Some of these panels are already completed, but hopefully the rest of the project will be added and completed gradually. The remaining projects include the recommender system project, the echo state network and reservoir computing package, the set of computer vision projects, etc. A link to download section will also be provided as I intend to publish all source codes and packages.

Data and Visual Analytics

D3 Visualizations

Using D3.js to visualize data; binding data to DOMs; applying data-driven transformations

Visualization Tweeter Graph

Retrieving twitter information using Tweepy, visualizing, and analyzing the connection graph

Machine Learning and Deep Learning

House Price Prediction Web Application

Machine learning-based interactive price estimation and visualization tool

Machine Learning Mini projects

A set of mini projects implementing machine learning concepts in Python and MATLAB

MapReduce on AWS and Azure HDInsight

Running a series of analysis on Google n-grams dataset using MapReduce techniques

Online Learning of User Privacy in Social Networks

A web plugin to evaluate the privacy risk and social benefit of sharing information in facebook

Deep Learning | CNN Projects

This page includes a set of multiple CNN-based computer vision projects that I worked on during Deep Learning courses or my own tests

Deep Learning | RNN Projects

This page includes a set of different RNN-based projects that their applications in processing sequential and temporal data

Deep Learning Recipes | RNNs, Echo State Networks

This page includes an introduction to echo state networks and their applications that I worked during my PhD research.

Time Series Forecasting | Autoencoder Echo State Network

Proposing a novel integrated approach, introduced as autoencoder ESN (AE-ESN), for long-term prediction of highly nonlinear time series. This architectur demonstrates huge improvements in the predcition accuracy and robustness.