Viewing profile — hahaker

Well, hold that thought, what if one-to-one means AI to AI...

Same here. What a confusing statement.

No you didn't recreated anything. That company is worth 500M. Your fun experiment is not. On the other hand, this is excellent PR for Opower.

Another huge market of AI is medical. e.g. algorithms are already doing a better job than human experts in detecting cancer.

Thanks for the comments! Are you asking the lyrics or music generation? For lyrics, we actually didn't train on Christmas songs. Training data was a large collection of romance nov…

I am the main developer. You are literally right on the time spent... Our original focus was writing a research paper on hierarchical music generation. Composing a song from an ima…

Very interesting reference! Deep learning is statistical, so this is sorta one of the spiritual predecessors.

I'm not very familiar with mad-libs so correct me if I was wrong. I think generating a lyrics passage (zero hard-coded rule on content or grammar or anything) from an image would n…

The Christmas element comes from 1. the image, and 2. a 4800-dimensional RNN sentence encoding bias generated from ~30 Christmas songs. Not sure how to hardcode this.

I'm from the project team. This is a very interesting point. While it is easy to crawl many songs from the internet, it is a little harder to gather the same amount but with proper…

This seems very useful, but isn't Google Maps doing similar stuff already? And if not, what are their concerns?

Good point

Obviously it is a convolutional neural net. Here you can find source code for one of the latest work: https://github.com/s9xie/hed