I need to have a management tool/environment driven by a Database and by content extracted from python files (specifically: tagged comments) while storing all relevant results of an “experiments” done with an neural network in a DB.
The goal is thereby to develop a (visual) interface from which I can see if the underlying code has matching interfaces (input, output, normalization function etc.) while using the referential integrity of a database model to ensure that changes made to the interfaces stay undetected.
Each run of the code should then also be considered as an “experiment” which is being documented via data within the DB – and would enable the user to get back to an earlier version/experiment and make some modification without overwriting previous code/settings.
The project was started by some other developer, who needed to exit the project because of personal reasons. You will get that project. The visual interface is being provided via sketches (together with an additional doc for more background on the proj.). I am expecting from the developer also to extract a database model from the requirements.
The entire project should be done in Python (3.6); but there is no need to do the visual interfaces in that language as well (if it can be done easier with other tools/languages, then I have no objections). Actually, I don’t care about visually pleasing interfaces, I am much more interested about functionality that is making me more productive.
It would be good to have a programmer who also has some experience with Neural Networks (TensorFlow or Keras) and know how to interface with these frameworks. Then I am sure you can contribute with some additional input to this project. However, a toy NN is already being provided with full code.
The first milestone should be to develop the management interface for that toy NN.
The second milestone is to apply the project to an application of more complex/structured input data. If the work done within the first milestone is being done well, then this second milestone should be realized really easily.
Focus should also be on providing an relatively easy & efficient (reusable/template kind of) code that helps me to manage, input, output and multiple hidden layer and the connection between them … I would expect that numpy, panda etc. is being used.
1. The signing of an “Intellectual Property Transfer Agreement” will be required.
2. I have already an Architecture and an Interface that I would share with qualified developers only.