PhD Student in Information Engineering · University of Padova

Spiking neural networks, local learning, and adaptive computation

I work on machine learning systems shaped by temporal dynamics, sparse activity, and biologically inspired adaptation, with a particular focus on spiking neural networks and neuromorphic learning.

Portrait of Aidin Attar

Research interests

Spiking Neural Networks Neuromorphic Computing Biologically Inspired Learning Energy-Aware Intelligence Continual and Adaptive Learning Machine Learning Systems Dynamical Systems Physics Inspired Models

About

I am a PhD student at the Department of Information Engineering of the University of Padova, where I work on spiking neural networks, local learning rules, and adaptive machine learning systems. My research focuses on how temporal computation, sparse activity, and biologically inspired plasticity can support more efficient and flexible forms of intelligence.

Before starting my PhD, I studied Physics and worked on machine learning problems in both academic and applied settings. Over time, my interests have moved toward questions at the intersection of machine learning, neuromorphic computing, and dynamical systems: how to train deeper spiking systems, how to combine fast adaptation with longer-term memory, and how learning can remain effective under resource and energy constraints.

I am especially interested in approaches that move beyond purely global optimization and instead exploit structure, dynamics, and local update mechanisms. Alongside theory and experiments, I also care about building research tooling that makes these ideas easier to test, reproduce, and extend.

Research directions

Main research direction

Deep spiking networks and local learning

I study how spiking neural networks can move beyond shallow pipelines through deeper architectures, structured connectivity, and learning rules such as STDP and reward-modulated plasticity.

Ongoing direction

Multi-timescale adaptation and memory

I am interested in systems that combine fast adaptation with slower consolidation, with the goal of improving continual learning, robustness, and online behavior under changing conditions.

Emerging direction

Predictive and energy-aware intelligence

I explore connections between spiking computation, predictive processing, and adaptive control in resource-constrained systems where sensing, learning, and computation must be carefully regulated.

Read more about my research

Selected projects

Sensing and machine learning

Human Activity Recognition with mmWave Radar

Activity recognition from mmWave radar data, combining signal processing and machine learning on real-world motion measurements across multiple subjects and environments.

View repository

Efficient audio models

Small-Footprint Keyword Spotting

A compact keyword spotting pipeline based on convolutional neural networks, designed for efficient audio classification in constrained settings.

View repository

Automation and tooling

Lab Automation for Shared Research Spaces

An automation tool for everyday lab workflows, built to support practical coordination and lightweight infrastructure tasks in a shared research environment.

View repository

Browse all projects

Writing

I am gradually building a public writing section for shorter research notes, conceptual clarifications, and implementation-oriented essays. Topics will likely include spiking neural networks, predictive coding, multi-timescale learning, local plasticity, and research software design.

Visit the notes page

Contact

The easiest way to reach me is by email. You can also find my code and technical work on GitHub.

Contact me