ims = data[4] # The arrays with the images
ims = np.transpose(ims)
ims = ims / 2.0 # Scale
ims = ims - 0.5 # Center
ims2 = ims**2
Nside = ims.shape[2]
frame_times = data[0] # The times at which the frames start
diff_frame = np.diff(frame_times, axis=0) # This calculate the duration of each frame
vm = data[2] # Extracts voltage
## Pre-process the signal
# sampling interval
sampling_time_interval = 0.102 # Obtained from Cyrills, experimental value
factor = 10 #
vm = downsample(vm, factor) # Sample down the signal by factor
first_image = int( frame_times[0] / (factor * sampling_time_interval)) # time of the first image
last_image = int( (frame_times[-1] + np.mean(diff_frame)) / (factor * sampling_time_interval))
last_image = int( (frame_times[-1] ) / (factor * sampling_time_interval)) # time of the last image
vm = vm[first_image:last_image] # Takes only the voltage that corresponds to images
################
## Data parameters
################
#Scale and size values
dt = sampling_time_interval * factor # Sampling time interval (ms)
dim = np.mean(diff_frame) # Duration of each image (ms)
ims = np.transpose(ims)
ims = ims / 2.0 # Scale
ims = ims - 0.5 # Center
ims2 = ims**2
Nside = ims.shape[2]
frame_times = data[0] # The times at which the frames start
diff_frame = np.diff(frame_times,
axis=0) # This calculate the duration of each frame
vm = data[2] # Extracts voltage
## Pre-process the signal
# sampling interval
sampling_time_interval = 0.102 # Obtained from Cyrills, experimental value
factor = 10 #
vm = downsample(vm, factor) # Sample down the signal by factor
first_image = int(
frame_times[0] /
(factor * sampling_time_interval)) # time of the first image
last_image = int((frame_times[-1] + np.mean(diff_frame)) /
(factor * sampling_time_interval))
last_image = int(
(frame_times[-1]) /
(factor * sampling_time_interval)) # time of the last image
vm = vm[first_image:
last_image] # Takes only the voltage that corresponds to images
################
## Data parameters
#Scale and size values dt = 1.0 # time sampling (ms) dim = 21.0 # duration of the image (ms) dh = 7.0 # resolution of the kernel (ms) kernel_duration = 150 # ms kernel_size = int(kernel_duration / dh) # Scale factors input_to_image = dt / dim # Transforms input to image kernel_to_input = dh / dt # Transforms kernel to input image_to_input = dim / dt # transforms imagen to input ## Input preprocesing vm = downsample(vm, dt) # Take the percentage of the total that is going to be used percentage = 0.30 Ntotal = int(percentage * vm.size) # Take the minimum between the maximum and the choice Ntotal = np.min((Ntotal, vm.size)) V = vm[0:int(Ntotal)] vm = None # Liberate memory # Size of the training set as a percentage of the data alpha = 0.95 # training vs total Ntraining = int(alpha * Ntotal) # Construct the set of indexes (training, test, working)
# Scale and size values dt = 1.0 # time sampling (ms) dim = 21.0 # duration of the image (ms) dh = 7.0 # resolution of the kernel (ms) kernel_duration = 150 # ms kernel_size = int(kernel_duration / dh) # Scale factors input_to_image = dt / dim # Transforms input to image kernel_to_input = dh / dt # Transforms kernel to input image_to_input = dim / dt # transforms imagen to input ## Input preprocesing vm = downsample(vm, dt) # Take the percentage of the total that is going to be used percentage = 0.30 Ntotal = int(percentage * vm.size) # Take the minimum between the maximum and the choice Ntotal = np.min((Ntotal, vm.size)) V = vm[0 : int(Ntotal)] vm = None # Liberate memory # Size of the training set as a percentage of the data alpha = 0.95 # training vs total Ntraining = int(alpha * Ntotal) # Construct the set of indexes (training, test, working)