Sudarshan Paul's Projects
The repo deals with an more advanced version of Recommendation Systems which is based on two different popularly known filtering mechanisms which are the Content Based Filtering and Collaborative Based Filtering. In Content based Filtering the main focus is on the item characteristics and features which are used to predict other items of similar nature. The main idea behind Collaborative based filtering is that we are given a matrix of preferences by users for items, and these are used to predict missing preferences and recommend items with high predictions.
The purpose of this following repository is to showcase the tasks performed by Sudarshan Paul during the AIChamp Marathon conducted by CodeVector Labs during 24/08/2020 to 28/08/2020.
The following repository contains the weekly assignments and Tasks performed by Sudarshan Paul in association with AI-Champ Program conducted by Code-Vector Labs.
This simple python project helps you in keeping a track of the price of your favorite items on any any e-commerce (Amazon in our case). If you ever wanted to automate the process of daily checking the price of any item whether it dropped on certain days, this app serves this purpose. It not only keeps a track of the price but also alerts you via mail when the price falls below a certain predefined (by you) value. So you can quickly grab it before anyone else does!
Customer Segmentation refers to the process of identifying several segments of customers from a pool of customer database which helps businesses to target potential user base. The technique of customer segmentation depends upon various differentiating factors such as demographics, economy status, geography etc. All these factors plays a vital role in the customer segmentation process. In order to perform this project, clustering mechanism of Unsupervised Machine Learning Algorithm is used specifically K-Means Clustering.
The aim of this project is to develop a Machine Learning model to predict the employees in a firm who are most likely to be promoted based on their previous performance and other various company specific attributes.
Image Compression is a technique in which the existing pixel size of a picture is reduced by adopting the most suitable least value. This least size of pixels in an image is determined by implementing the use case of K-Means clustering algorithm. K-Means is an unsupervised machine learning algorithm in which the un-labelled data points are sorted to form k clusters based on certain similarity criterion such as euclidean distance between them. The value of K is usually provided by the user and depends upon certain factors. In this project the value of K will determine the total number of colors that will be used to reproduce the image, as expected the value of K will certainly be less than the total number of colors used in the original image(which is in fact very large).
The project on Language Classification using Naive-Bayes algorithm deals with classifying as well as identifying the language of input string into its correct category. In order to demonstrate the use-case i have trained the model to detect three different languages with an adequate accuracy. The languages that i have chosen for this model are Slovak (sk), Czec (cs) and English (en). The purpose for choosing slovak and czec is that both of these languages are very similar in the way they are spoken, so if the model is able to distinguish between these two given languages that will ensure the robustness of our model to classify between other languages with a very good accuracy.
MLOps hands-on notes, notebooks and scripts of months of learning
The following project deals with a simplified version of movie recommendation system which are deployed in various web services these days from your favorite Netflix, Amazon Prime etc. The effort has been made to implement the basic idea behind these recommendation systems.
The repository contains configuration files for various deep learning models served using triton server.
The mentioned project deals with classifying a bunch of messages obtained from different sources into spam or ham by using basic Natural Language Techniques and Naive-Bayes Classification algorithm. Basically the working of this model can be summarized as it trains on a set of cleaned training data and tested upon an unseen set of data or Test data which contains random labelled messages as being spam or ham. The output of this model is that the message either being classified as spam (1) or ham (0).
The following repository has been created to showcase the tasks performed during the "The Sparks Foundation - Machine Learning /Data Science" Graduate Rotational Internship Program July-August.
GitHub repo for Azure Video Analyzer
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent