Rama Natarajan - Courses

Research     Courses     Teaching

Reports     Misc     Home

Introduction to Machine Learning - Fall 2004

Project: Segmenting and Labeling Rhythmic Multi-unit Neural Response using CRFs, Rama Natarajan.
The paper presents details of a log-linear sequence labeling model that is trained discriminatively with a general-purpose optimization method, as a simple yet competitive solution to detecting bursts of neural activity in neurophysiological recordings. Experimental results on neural burst extraction and labeling tasks on recording from crustacean feeding system are presented. Implementation: Code | Results | Documentation

Computational Neuroscience - Fall 2003

Course: Overview :: Instructor: Dr. Richard Zemel :: Course Webpage

Text: Theoretical Neuroscience, Computational and Mathematical Modeling of Neural Systems by Peter Dayan and LF Abbott

Team Presentation: On Working Models of Memory. Rosalia Christodoulopoulou, Horst Samulowitz and Rama Natarajan

Project: An Implementation of the Extended Poisson Model for Distributional Population Codes, Rama Natarajan and Horst Samulowitz.
Abstract: We examine the Extended Poisson Model for encoding and decoding probability distributions in populations of neurons. We derive the equations for two MAP Estimation methods: Expectation- Maximization and gradient descent using Softmax function, to reconstruct an input distribution based on the observed firing activity of the neurons. We analyze the performance of these decoding methods in extracting information regarding multiple values and uncertainty in the value of the stimulus variable encoded, and demonstrate that it is robust to noise. We also examine the effect of smoothness priors and weighting factors used in the model, in the convergence rate of the algorithms and quality of reconstructed distribution. Implementation: Code | Results | Documentation

Neuroscience: Systems and Behaviour

Course: Overview

Text: E.R. Kandel, J.H. Schwartz and T.M. Jessell: Principles of Neural Science (Fourth Edition). McGraw-Hill, 2000.

Project: Cellular and Molecular Mechanisms Underlying Fear Memory Consolidation in the Amygdala: What Triggers CREB Activation?

Abstract: Studies of implicit memory of fear in mammals have shown that long-term memory consolidation requires highly co-ordinated cellular and molecular changes. These changes involve alterations in synaptic transmission, and a cascade of gene expression and protein synthesis. Fear learning is facilitated by synaptic plasticity in the amygdala. The transcriptional cascade that controls the formation and stability of long-term fear memory is believed to be triggered by the nuclear transcriptional factor cAMP response element-binding protein (CREB). This paper reviews recent findings on mechanisms of calcium influx that could trigger the cascade of intracellular processes, and the role of intracellular signaling pathways in conditioning to auditory and contextual stimuli. PKA, ERK/MAPK, CaMK, CREB, RNA, and protein synthesis in the amygdala have all been shown to be necessary for the formation of long-term fear memories. The review attempts to elucidate the cellular mechanisms that lead to the activation of these protein kinases and subsequent molecular changes which trigger the activation of CREB.