Product recommendation system

Using a russian online marketplace company Avito’s dataset containing features like: UserID ProductID AdID UserDemoraphics AdPosition Click …. and many more. The main aim was to produce personalized product recommendation based on the user interaction with products and similarity with other users. I have tried to prevent any monotonus recommendations by keep the accuracy low to include different products. LightFM model was used to include both user and Ad meta deta to determine the same.

April 11, 2021

Face Mask Detection

Aim of the project was to build a object detection model to identify whether a person is wearing a mask or not. In the time of this ongoing pandemic, we are forced to wear mask whenever we go outside, hence this can be used by various offline stores or offices to be aware that whether a customer/employee is entering without a mask and hence preventive measures can be taken. Model is yet to be tuned as the framerate is slow and hence need extra resources to make it work in realtime.

May 10, 2020

Fake news ML service

Overview People are daily bombarded by various types of new from different resources and sometimes some of them are fake. These fake news can lead to various unpleasant events and hence has to be stopped. This is an practice approach to process news and classify them using Machine Learning into fake or real. Workflow All the news data are extracted using the data_acquire script which are labelled. This data is then processed using NLP techniques like stemming, lemmatization and conversion into vectors for our models to understand them.