I actually had a similar idea that I cam across in my own research in brain-to-video applications. You could theoretically train the model in real-time by using brain data as a reward signal, such that the model would find the image or video that most strongly produces the desired response (e.g. an increase in certain brainwave bands associated with pleasure, or alertness, or fear).
I can imagine this has some very undesirable usecases if used in the wrong way.
I've been working on my own GAN for a while, trained on a few different datasets. Then the EEG data is fed into the GAN in real-time and used to interpolate within it's latent space. I assign each brainwave band (e.g. alpha, beta, etc.) to it's own point within that latent space, so the core idea is that as alpha increases (for example) the output heads towards that point in the space and that's reflected in the output image. Happy to talk more.
Using a Muse EEG headset to read brain activity and use that to drive the output of a GAN. Similar to other projects that try to visualise or decode thoughts, but at the moment it's an art project. Obviously quite limited by compute and hardware. I'm sort of looking for collaborators / co-founders / opportunities in the AI + neuroscience + creativity space.
Using a Muse EEG headset to read brain activity and use that to drive the output of a GAN. Similar to other projects that try to visualise or decode thoughts, but at the moment it's an art project. Obviosuly quite limited by compute and hardware. I'm sort of looking for collaborators / co-founders / opportunities in the AI + neuroscience + creativity space.
I'm sort of thinking out loud here but could you have two batteries running simultaneously but on opposite cycles, so while one is cooling the other is heating? Obviously it wouldn't be 100% efficient but it might reduce some wasted energy.