-
Notifications
You must be signed in to change notification settings - Fork 4
Troubleshooting
In my data, the time step changes from file to file, and this causes the neural network to fail. How do I sort spikes?
Use the global network. The global neural net works for any time step in your data
This might be due to the presence of a corrupted Matlab file from trying to generate a previous network, or sometimes something goes wrong and the neural network file gets corrupted in another way. The neural network folder isn’t stored in the same place as the rest of the crabsort files (crabsort-->crabsort-master-->spikes-->notebook_exp#) so deleting this folder doesn’t solve the issue and this will keep happening when you start over. The neural network file Matlab generates is stored in the folder where the raw data files are located in a network folder (notebook_exp#-->network-->ch_name). In this network folder there is a folder for each channel with the file for the neural network generated for that channel. Deleting the specific corrupt channel folder with its associated files will fix the problem.
Don’t resize the window when you first open crabsort, otherwise this dropdown menu can disappear on smaller screen sizes! It’s usually best to label all the channels right away before you do anything else.
Under the “Tools” menu, click “Auto Y Zoom”
Under the “Tools” menu, click “Reset Current Channel,” this will delete any spike information you did on this channel for this file, but won’t delete any automate data, or spike data on other files.
Don’t worry, you didn’t lose everything since crabsort saves as you go along. You may lose a little bit depending on what stage you were at and when crabsort last saved, but not everything. Reopen Matlab and then crabsort. Open the last file you were working on. If this file is one you were predicting spikes in after generating a neural network, you will need to hit “p” to predict the spikes again. If you were prior to generating a neural network, crabsort should load with the saved work you’ve done. If you were actively generating a neural network, crabsort may or may not have saved the generated network.