Possible other uses of generative networks

Generative networks seem oriented to language, chatboots, writing, reading, etc.

Are they appropriate for other types of work?
Industry, generation of medicines,…

1 Like

Oh wow! @Fernando_J_Zamora there are so many uses of Generative networks.
out of the box, and most commonly known, is for image generation (like paint like V. van Gogh style. I am not an expert in the use in music composition, but it could be also used for that.

What you call generation of medicines, its name is Drug Discovery (I am sure you will find plenty if you Google that term), to my understanding is about predicting molecular structures and binding. This is used a lot by the way… by the pharma industry, it saves so much time and money.

Audio Denoising! this is a good one!! and also made famous by Nvidia, but also now in many Phone devices. Also consider that Autoenconders can be considered Generative. It comes to mind VAEs. You can go on, for image denoising… etc.

Voice generation… there are many ways to do this, but one is using this type of networks, but your problems mostly lie on data, and labelling.

I dont want to bore you but, if you are into small bots, or RC cars, VAEs have been also used to train… GitHub - araffin/learning-to-drive-in-5-minutes: Implementation of reinforcement learning approach to make a car learn to drive smoothly in minutes

Enjoy! :tada:

2 Likes

I’ve made a few ‘syntherhals’ one of Jorges Luis Borges and the other Rick Roderick, both deceased individuals whose work can live on via AI being trained on their works.

I am wondering if LLMs can be used to create a rule based language model. Once we get to the rule based model, i can possibly be used for specific domain based queries where we can provide the domain knowledge base. We can thus prevent hallucination and generation of information that is not fact based.

1 Like