Q1)
- Run the script file naming l31script.m . The training error is shown on matlab command window after each epoch, calculated as mentioned in question.
- You will get 7 different graphs in which 6 graphs are of decision boundary with H = 2,4,8,16,32 and 64.
- One graph will be of training error for differnt H values.
- Combining all the training errors in one graph. The final graph is shown in the report.
Q2) Task1
- To accomplish this task, run the file naming script2.m . The training and validation error is shown on matlab command window after each epoch.
- All the images of the training set and validation set should be in the folder naming 'steering'.
- Set all the values in the script file to get differnt plots.
- Set Number of Epochs, learning rate, minibatch size, dropout probability of all the layers.
- Now run the file script2.m and get the desired plots. (*however the results obtained during the run are listed in Results.txt)
Q3) Task2
- For this run the file naming script2comp.m.
- All the images of the test set should be in the folder naming 'l3-test'.
- Get the file 'ydash.mat' which shows the predicted steering angle for the corresponding image. (* the training weights are saved in the files namely savew1.mat, savew2.mat, savew3.mat)
P.S. Make sure that all the files are in the same folder.
Directory structure (2015csb1038_2015csb1014_lab3.zip)
|
|
1. Code files
|
|
1. .m files - all functions and script files
2. .mat files
3. README.txt
4. Results.txt
2. l3-test
|
|
1. Images of the test dataset.
3. steering
|
|
1. Images of the training and validation dataset.
4. Q_1_plots
|
|
1. Plots for the question 1.
5. Q2_a_plots
|
|
1. Plots for the question 2-a.
Thank You