Master's Thesis

Ce mémoire de Master explore la conception de circuits analogiques pour le calcul neuromorphique. Une opportunité unique d'intégrer l'innovation dans une équipe dynamique.

Tâches

• Développer des modèles de neurones à impulsions en utilisant des techniques analogiques.

• Tester la scalabilité des circuits dans un petit réseau neuronal.

• Utiliser la technologie CMOS avancée pour le design de circuits.

Compétences

• Diplôme en génie électrique et compétences en conception de circuits analogiques.

• Connaissance des amplificateurs opérationnels et de Verilog.

• Compétences en Python et utilisation de Cadence Design Suite.

Master's Thesis
Analog Circuit Design for Neuromorphic Computing
Ref. 2025_013
Neuromorphic computing is a computing approach inspired by the structure and function of the human brain. This work focuses on spiking neuron models, where communication between neurons occurs via temporally encoded spikes. These are short pulses emitted by the transmitter (TX) of one neuron. These pulses are then received by the receiver (RX) of the next neuron and integrated over time. The transmitted information is extracted from the integrated value.

In previous work, the described communication link was implemented using digital integrators. However, this approach requires an underlying, fine-grained clocking scheme that consumes a lot of power.

In this work, we aim to overcome this disadvantage by implementing the neuron transceiver (TX, RX) using analog circuit techniques. Important building blocks may include analog integrators, summers, or even multipliers.

The goal is to test this analog circuit approach in a small neural network and evaluate its scalability to larger neural networks. The circuit design is done using leading-edge CMOS technology.
Requirements
• Bachelor’s degree in electrical engineering.
• Knowledge of analog circuit design (e.g., operational amplifiers) and Veriloga skills (for behavioral modeling of analog components).
• Knowledge of Cadence Design Suite (Circuit Simulator) and basic Python programming skills for neuron modeling in frameworks (PyTorch, TensorFlow) are advantageous.

The Master Thesis will take approximately 4 months to complete. Ideally, interested candidates should have a professor at their university as their contact person for supervision. On-site work is desirable. Exceptions for remote work are possible. IBM does not offer financial support for this Master thesis.

This position is available starting immediately or at a later date.
Diversity
IBM is committed to diversity at the workplace. With us you will find an open, multicultural environment. Excellent flexible working arrangements enable all genders to strike the desired balance between their professional development and their personal lives.

How to apply
Please submit your application through the link below.

If you have any question related to this position, please contact Dr. Marcel Kossel, E-Mail schreiben.