ELEC 631 - Advanced Digital Signal Processing: Signal Processing, Information Theory, and Machine Learning Methods for Neuro-Engineering - Spring 2012

---

This course will explore how mathematical tools from signal processing, information theory, and machine learning have been applied in neuro-engineering.

Topics to be discussed include: Information coding and neural activity; Correlated activity and neural circuits; General place cell/grid cell and navigation; Theta phase precession; Grid cells/place cells (oscillatory interference); Poisson processes and neural decoding (and adaptive filters); Mathematical model of learning behavior; Spectral estimation; What is the unit of neural computation?; Inferring spikes from calcium-sensitive fluorescence imaging; Subresolution microscopy; Spike sorting; Prosthetic decoding algorithms.

---

Instructors:

Location: Duncan Hall 1044

Time: Friday 2-4pm

This is a "reading course", meaning that students will read classic and recent papers and present to the rest of the class in a debate format. Students will also complete a group project. The course is open to graduate students from any department with some background in signal processing, information theory, and machine learning.

Class grade will be based on:

• class participation (20%)
• paper presentations (30%) - SCHEDULE
• paper summaries (prepared each week, 10%)
• group project (40%)

Additional Resources

• Computational Explorations in Cognitive Neuroscience (O'Reilly & Munakata, 2000)
  
  

Weekly Schedule

Model-based decoding: how can we model neurons and optimally extract the information they represent?

(Jan 13) Introduction. Neurons, Rate coding, Sensory tuning, Motor tuning, Population coding.

• Hubel and Wiesel (1962)
• Neuronal Population Coding of Movement Direction, Georgopoulos et al, 1986.
  

(Jan 20) Optimizing system design for neural decoding

• A high-performance brain–computer interface, Santhanam et al., 2006.

(Jan 27) Linear neural coding for hand movement parameters

• Spatiotemporal Tuning of Motor Cortical Neurons for Hand Position and Velocity, Paninksi et al., 2004

(Feb 3) Hippocampal neural codes for space and a prior model for movement

• A Statistical Paradigm for Neural Spike Train Decoding Applied to Position
  Prediction from Ensemble Firing Patterns of Rat Hippocampal Place Cells, Brown et al., 1998

(Feb 10) What happens if the neural code for space changes?

• An analysis of neural receptive field plasticity by point process adaptive filtering, Brown et al 2001

(Feb 17) What happens if the neural code is state dependent?

• Neural Decoding of Hand Motion Using a Linear State-Space Model With Hidden States, Wu et al., 2009

(Feb 24) No Class

(Mar 2) Spring Break

Population Coding: What principles underlie neural codes? How is information represented across a population of neurons? How do neural systems operate?

(Mar 9) Covariance and population coding - Is it better to have independent neurons or correlated ones?

• Spatiotemporal correlations and visual signaling in complete neuronal population, Pillow et al., 2008

(Mar 16) Information representation in the brain - scaling, accuracy, what is ideal and what is observed?

• Grid cells generate an analog error-correcting code for singularly precise neural computation, Sreenivasan et al., 2011

(Mar 23) What are the principles of motor planning, how are the details of movements generated?)

• Signal-dependent noise determines motor planning, Wolpert, 1998

(Mar 30) Competition, Inhibition, and Cooperation amongst Population

• Locally Competitive Algorithms. Rozell et al., 2008
• Supplemental Material:
  

• Competition and cooperation in neural nets. Amari et al., 1977.
• Computing with neural circuits: a model, Hopfield, 1986.

Methods/Hardware for Brain Machine Interfaces (BMI): Challenges in algorithm design and hardware development for BMI.

(Apr 6) Guest lecture: We Ji Ma, Neuroscience Dept., BCM

(Apr 13) Electrophysiology and spike sorting

• Dimensionality reduction / clustering algorithms
• Hardware challenges: lower power, wireless
• Spike adaptation during bursts, colored noise
• Dynamic bit rates, compressive sensing, finite-rate-of-innovation