Exercise – spike train analysis The directory /DATA/ contains data from neurons recorded during an eye movement direction tuning task. Each file contains the information from a single neuron organized...

When can I have this exercise done and how much will it cost me? Is this a reliable site?


Exercise – spike train analysis The directory /DATA/ contains data from neurons recorded during an eye movement direction tuning task. Each file contains the information from a single neuron organized in a matlab structure. The name of the structure is data. It contains the following fields. 1. Spikes: T X N sparse matrix, N is the number of trial and T is time in ms. The matrix indicates times of the spikes in each trial. A value of 1 in indicates that the neuron had an action potential in the corresponding bin. 2. target_direction: a vector containing the direction in degrees of the target motion. 3. target_motion: The time of target motion onset and offset in ms. Questions (50% of gade) Please submit graphs with a very short explanation All answers in this part should use the data in the file named “cell3247.mat”. 1. Write a matlab script that shows the raster for each target motion direction separately. Hints: matlab function spy. 2. For each direction calculate the PSTH in 1 ms bin. Smooth the PSTH with a square window of 100 ms. Normalize the PSTH to firing rate in units of spikes/s. Present all 8 PSTH in a single matlab plots. Be sure the presented traces do not contain edge effects. Hints: matlab function smooth. 3. Calculate the direction tuning of the neuron. Calculate the average firing rate across all the movement time as a function of the direction of movement. Present a plot of the average firing rate across trials as a function of motion direction. 4. Fit the direction tuning to a cosine function: a. Present the fit together with the direction tuning of the cell. Use a resolution of 1 degree to present the results of the fit. b. Use the fit to estimate the preferred direction of the neuron. c. Bonus: firing rate cannot drop below 0, add this constraint to the fit. Find in the data set a cell in which this makes a difference and present the fit together with the real firing rate. Hint: Matlab functions fittype and fit. Main project (50% of Exercise grade + 5 bonus points to final grade) Please submit: 1. The matlab function by e-mail to yyyyy name the function classifier_[your name].m 2. A short explanation (up to half a word page) about the algorithm you used. The goal of this part of the exercise is to decode the direction of target motion from the spike train. You are required to write a classifier that receives as input a training set and a single spike train to classify. The classifier should use the information from the training set to predict the direction (in degrees) of motion in the test spike train. We provided a few functions that might help you in the process of testing your classifier: · [success _angle, success _id] = test_classifier (classifier_func, data) The inputs to the function are the classifier function you have written and the data from a single neuron. The function uses leave-one-out method to test the quality of the classifier and returns two values: 1. success_angle: the average angle difference between the classification output of classifier_func and the actual data. 2. success _id: The fraction of trials in which the classifier has correctly identified the motion direction. · classifier_random.m You could test the function test_classifier by running a classifier that randomly assign a direction to a spike train. The random classifier is implanted in the file classifier_random.m. Load data from one of the neurons then run the command: [success_angle, success_id] = test_classifier(@classifier_random, data). If everything is working fine you should get values around 90 for success_angle and 0.11 for success _id. · run_test.m You could (if you wish) use the script run_test.m to test your classifier on all the neurons. Your goal is to maximize the success_id and minimize the success_angle. To test your classifier insert it to a directory named /classifiers/ then add the function to the list of classifiers in the variable named “class_func” (~line 9 in the code). After you submit you code I will test your classifiers. Classifiers that will be better than a classifier I wrote in both measures will add a bonus of 5 points to final course grade.