Azal Amer

My research:

I am developing convolutional neural networks (CNNs) to distinguish ALPs from background photons using calorimeter simulation data. My work focuses on optimizing sparse convolution architectures and multi-task learning approaches to improve classification performance across different mass ranges.

My expertise is:

Convolutional Neural Networks

A problem I’m grappling with:

Identifying why improving the distinguishibility in the spectrum of features reduces the performance of a my model to classify.

I’ve got my eyes on:

Improving the readability and documentation surrounding my work, along with revisiting some of my older tests on newer code.

I want to know more about:

Using existing tools to interpret information about the features neural networks I train, extract.