In Neuromorphic Engineering very-large-scale integration (VLSI) systems containing analog circuits are used to create biological inspired structures such as those found in the nervous system. We can offer projects for both simulation of neural networks and design/evaluation of VLSI circuits which cover the following topics:
- Detection of spiking activity of biological neurons
You will design and characterize an analog VLSI circuit capable of detecting spikes within hippocampal slices of mice based on recorded signals from a multi-electrode array.
- Learning in memristive systems
Based on models obtained from physical devices you will simulate neural networks consisting of memristive devices and integrate-and-fire neurons. Different network topologies and learning performances will be evaluated on common machine learning tasks.
- Recurrent Competitive Networks
Formation of attractors in competitive neural networks will be examined. Therefore, the robustness and strength of an attractor neural network learned on a reconfigurable analog neuromorphic chip will be investigated.
- Response characterization of the spiking elementary motion detector
You will characterize the spiking elementary motion detector's response with respect to contrast and illumination, and compare it with biological data from the literature.
- Neuromorphic tempotron applied to speech recognition
You will use the SpiNNaker board to simulate a tempotron and apply it to keyword spotting. You will thoroughly characterize the system's performances.
- Perceptual decision in a recurrent neural network
You will use the SpiNNaker board to simulate a winner-take-all-network for performing motion based perceptual decision. You will thoroughly characterize the system's performances.
- Self-motion perception
You will use the SpiNNaker board to compute self-motion using a spiking elementary motion detector. You will thoroughly characterize the system's performances.
- Power-law adaptation
You will design and characterize an analog circuit for implementing
power-law adaptation in a spiking neuron. Experience in Cadence or
passed Neuromorphic Engineering II is a prerequisite for working on this
- Local vs. global competition in latency coding
You will characterize and compare contrast enhancement mechanisms based
on local and global inhibition in a network of spiking neurons.
- Implementing learning in sensory fibers for dynamic analyses of static images
You will explore and characterize learning mechanisms in a spiking neuron
model of sensory fibers based on biological evidence.