Mahendran Narayanan

Published Machine Learning Researcher with 2 peer-reviewed ( ICPR 2024 , IntelliSys 2024) and 2+ years of applied experience. Proven ability to bridge the gap between theory and practice, with a focus on image classification, computer vision, and natural language processing. Skilled in developing and optimizing state-of-the-art models, including CNNs, vision transformers, and language models. Proficient in cloud computing (AWS, docker, containers). Also have research experience in cutting-edge techniques like sparse matrix multiplication for model acceleration, Quantization pipelines testing.

Areas worked: Network engineering, Vision Transformers, Image classification, machine unlearning, Finetuning LLMs, Generative AI, Regularization techniques (DL), Hyperparameter tuning, Model inference Testing.
Areas open to: Deep learning, LLMs, Generative AI, Computer vision.
News

2024 Paper accepted in ICPR 2024 conference
2024 Paper accepted in IntelliSys 2024 conference
Research Google scholar
SENetV2: Aggregated dense layer for channelwise and global representations [ IntelliSys 2024 ]
Proposed a aggregated dense layer within the Squeeze-and-excitation module of a neural network.
This enhances learning representation in SENet.
Variations of Squeeze and Excitation networks [ ArXiv ]
Proposed five variants of Squeeze and excitation module.
This change in module structure impacts the accuracy to a certain extent in SEResnet.
Deep learning for Fitness [ ArXiv ]
Application paper to compare and verify the body posture in real-time with a reference image and provide immediate feedback and suggestions.
Handwritten stroke augmentation on images [ CVPR 2022 (Rejected)]
Authored a novel OCR based universal augmentation technique for OCR texts irrespective of language.
Memory visualization framework for training neural network [ ArXiv ]
Created a web based UI for monitoring memory consumption w.r.t hyperparameters in network.
Identifying tourist destinations from movie scenes using Deep Learning [ ArXiv ]
Developed a methodology along with a curated dataset for identifying travel destinations depicted in movies based on the scenic elements portrayed.
Project Experience

Generative AI Replace backgrounds Code
Replacing the background of the generated image.
Used Stable Diffusion (SD) model and Segment Anything (SAM) model to perform background replacement on the generated image.
GPT unlearning Code
Simple try to make GPT unlearn alphabets
Generated dataset by GPT helps in making GPT unlearn Alphabets.
Outcomes states "A,B,C no longer represent alphabets by the resultant GPT model." Note : Model is heavily finetuned.
Tamil MNIST dataset
Collected a dataset consisting of more than 36,000 samples for Handwritten OCR specifically tailored for the Tamil language.
Replicating papers
Replicating network engineering papers within the domains of image classification and LLMs.
Experimenting with the latest modules and open-sourcing the project on GitHub.