Master’s student at SCU in CS, passionate about creating user-centric applications with AI and ML integration.

medium

github

Siddhant Kandge  • Peerlist

The project used MIT's algorithm to debias facial detection systems
using generative models. The goal was to reduce biases during image classification. A
dataset was created, cleaned, normalized, and trained on various architectures.We used
hyper parameter tuning to achieve robust results. In the end, we published our work to
arxiv.org - a preprint server and open repository of Cornell University.
Link to Publication : https://arxiv.org/abs/2204.09556

Debiasing Facial Detection System

This project works on Flicker8k dataset to generate caption to given image using pretrained encoder is used.(InceptionV3)
It is a model that generates sequences that uses an encoder to encode input
image to fixed compressed form and then decoder decodes it, word by word into
a sequence.

Image Captioning using TensorFlow

In this project, we implemented CNN based malware classifier which
takes files from the system as input and classify them as malicious or benign. For the
dataset, we took it from kaggle and on top of it, we made use of data augmentation to
generate more sample instances of data. This project, we created classifier architecture
and trained it on this dataset. Also, We used hyper-parameter tuning to achieve more
accurate end results.
The project can be found at https://gitlab.com/sign_lang_recg/malware-detection-app