Como classificar dados estruturados usando camadas de pré-processamento do Keras avaliações

Miguel Ángel H. · Revisado há almost 2 years

Himaja G. · Revisado há almost 2 years

Bharat C. · Revisado há almost 2 years

Ramakrishna P. · Revisado há almost 2 years

Banashree S. · Revisado há almost 2 years

Afrizal Ihyauddin Y. · Revisado há almost 2 years

Pranav S. · Revisado há almost 2 years

David C. · Revisado há almost 2 years

teo h. · Revisado há almost 2 years

Luis A. · Revisado há almost 2 years

Abid I. · Revisado há almost 2 years

Adriano O. · Revisado há almost 2 years

s. t. · Revisado há almost 2 years

Kelvin O. · Revisado há almost 2 years

Jayaraman S. · Revisado há almost 2 years

A consistent theme I've found with these trainings is a lack of comments within the code. For example in this lab in the "model.compile" part the "optimizer='adam'" is used. What is the adam optimizer? And what does it do? Why is it important? Of course I can google tf keras optimizer and find this - https://keras.io/api/optimizers/adam/. But why not include some of this information on optimizer in the code and lab? I think this would be useful for learning purposes. I really liked the keras.utils.plot_model that visually showed the model. This was great to include. Another recommendation I have is in the documentation of API, for example Keras API, functionalities could be described more simply. For example, what is an Epoch? I watched the TF videos leading up to this lab, and remember Epoch being mentioned, but did not fully absorb this information. So when I saw epoch in the lab I wondered, what is this doing? So I looked in the keras documentation (https://keras.io/getting_started/faq/) and this was the definition - "an arbitrary cutoff, generally defined as "one pass over the entire dataset", used to separate training into distinct phases, which is useful for logging and periodic evaluation.". On stackover flow someone described epochs in a more digestible and understandable way - "a number of epochs means how many times you go through your training set". Personally the stackoverflow explanation is easier to follow for me. For users to really learn I think there needs to be an emphasis on plainly and simply describing what various API's, functions etc. do. I know the stackoverflow response to what is an epoch is the layman's definition and not the computer science definition, but if Google wants a broader audience to learn their APIs and tools, then I would argue including a technical and simple explanation would help. Now that I have done a bit of critiquing, I will say this lab was very useful. Without this lab I would not know how to preprocess data for tf keras, build pipelines, fit a model and apply a model using TF keras. This has lab has the building blocks needed and workflow to apply keras to other data. I appreciate the work that went into the code. So a few more comments in the code and some more text explaining what various arguments within functions are doing would be very much appreciated.

Joshua H. · Revisado há almost 2 years

Scott W. · Revisado há almost 2 years

Sławomir M. · Revisado há almost 2 years

Juan Antonio H. · Revisado há almost 2 years

Mehdi T. · Revisado há almost 2 years

Fathu R. · Revisado há almost 2 years

Kharismadina H. · Revisado há almost 2 years

odj G. · Revisado há almost 2 years

Shota F. · Revisado há almost 2 years

Ojas M. · Revisado há almost 2 years