dnnbrain.io.fileio module

class dnnbrain.io.fileio.ActivationFile(fname)

Bases: object

a class to read and write activation file

read(dmask=None)

Read DNN activation

Parameters

dmask (dict) – Dictionary of the DNN mask information

Returns

activation – DNN activation

Return type

dict

write(activation)

Write DNN activation to a hdf5 file

Parameters

activation (dict) – DNN activation

class dnnbrain.io.fileio.MaskFile(fname)

Bases: object

a class to read and write dnn mask file

read()

Read DNN mask

Returns

dmask – Dictionary of the DNN mask information

Return type

OrderedDict

write(dmask)

Generate .dmask.csv

Parameters

dmask (dict) – Dictionary of the DNN mask information

class dnnbrain.io.fileio.RdmFile(fname)

Bases: object

A class to read/write representation distance matrices (RDMs) from/to .rdm.h5 file. For saving time and space, DNNBrain only hold on the upper triangle of each RDM. We can use np.tri(n_item, k=-1, dtype=np.bool).T to get the index array of the upper triangle. The index array can help us to restore RDM from the upper triangle whose shape is ((n_item^2-n_item)/2,).

read()

Read RDMs from .rdm.h5 file

Returns

• rdm_type (str) – Choices=(bRDM, dRDM) bRDM: RDM for brain activation dRDM: RDM for DNN activation

• rdm_dict (dict) – A dict containing the RDM information

  RDM type	Keys	Values
  bRDM	Labels of brain ROIs	Arrays with shape as ((n_item^2-n_item)/2,)
  dRDM	Layer names	Arrays with shape as (n_iter, (n_item^2-n_item)/2)

write(rdm_type, rdm_dict)

write RDMs to .rdm.h5 file

Parameters

• rdm_type (str) – Choices=(bRDM, dRDM) bRDM: RDM for brain activation dRDM: RDM for DNN activation

• rdm_dict (dict) –

  A dict containing the RDM information

  RDM type	Keys	Values
  bRDM	Labels of brain ROIs	Arrays with shape as ((n_item^2-n_item)/2,)
  dRDM	Layer names	Arrays with shape as (n_iter, (n_item^2-n_item)/2)

class dnnbrain.io.fileio.RoiFile(fname)

Bases: object

a class to read and write roi file

read(rois=None)

Read data of ROIs of the brain.

Parameters

rois (str, list) – ROI names of interest

Returns

• rois (list) – ROI names which are corresponding to columns in data

• data (array) – ROI data

write(rois, data)

Write data of ROIs of the brain.

Parameters

• rois (str, list) – ROI names which are corresponding to columns in data

• data (array) – ROI data

class dnnbrain.io.fileio.StimulusFile(fname)

Bases: object

A class to read and write stimulus file

read()

Returns

stimuli – Dictionary of the stimuli information

Return type

OrderedDict

write(type, path, data, **opt_meta)

Parameters

• type (str) – Stimulus type in (‘image’, ‘video’, ‘video_clip’)

• path (str) –

  Path_to_stimuli.

  If type is ‘image’, the path is the parent directory of the images. If type is ‘video’, the path is the file name of the video. If type is ‘video_clip’, the path is the parent directory of the video clips.

• data (dict) – Stimulus variable data

• opt_meta (dict) – Some other optional meta data