Deep Learning Framework using Big Data for tabular data (i.e. not image or sequence or time series data)

Deep Learning framework provided seems to support input layers pertaining to image and sequence/time-series data only. Is the understanding correct? Are there means to use for tabular non-sequence big data as input (via datastore tall arrays or any other equivalent means?) and appropriate intermediate layers and output layer? For instance, have a data store (/tall array) as input layer, followed by leakyReluLayers, and a regression layer output.

Answer 1

[Vishal Bhutani]

By my understanding you want to create a deep learning framework for non-sequence or non-image data, similar question related to this has been asked earlier. I am attaching it’s link hope it helps:

https://www.mathworks.com/matlabcentral/answers/395817-can-i-use-trainnetwork-to-train-deep-neural-networks-with-non-image-or-non-sequence-data-for-regre?s_tid=answers_rc1-1_p1_MLT

[Ramakrishnan Raman]

Thank you. Could you please clarify if the suggested step would work for Big Data? My understanding is that the ImageDatastore object would be able to handle that, for the image data case. In case of tabular data (non-sequence, non-image), how should this be done? The nExamples in the example provided in the link is ~a million