ML/AI Engineer | Hobbyist Apple Developer

Ankit Kumar  • Peerlist

- Collected the IMDb dataset for movie reviews.
- Pre-processed the dataset using tools from nltk library and applied concepts of NLP - Bag Of Words, Stemming, Removal of words and Conversion to feature vectors.
- Trained the model using Logistic Regression and made a web app using Flask (Python).
- Got 90% test accuracy.

Sentiment Analysis for IMDb Movie Reviews

Applied Convolutional Neural Network to predict and classify an image as a digit based on training.

Handwritten Digit Recognition

- Compared the results for number of neurons impacting performance of the hidden layers.

Study of Impact of hidden neurons to model performance.

- EDA on the titanic dataset provided in kaggle competition
- Analysed various correlation and factors for survival of onboard passengers.
- Applied Machine Learning to predict who may have survived and who may not have.

Titanic Survival - Machine Learning

- Used CNN to classify dog breeds using Keras library based on object localization
- Used concepts of Transfer Learning and implemented the ResNet-50 state of the art model for our data to better the results

Dog Breed Classifier

Applied Multiple Linear Regression technique to predict the price of used cars based on the distance they have been driven.