3 ms·
We call it deep learning, but it is deep pattern matching. Extremely useful, but don't expect it to result in thinking machines.
by ozy 8y ago
We call it deep learning, but it is deep pattern matching. Extremely useful, but don't expect it to result in thinking machines.
- halflings 8y agoAre our brains magic? If they aren't then surely they must be doing something that we can reproduce. We've built so many things that we considered "thinking machines" in the recent past (realistic speech synthesis, image recognition and captioning, human-level translation, elaborate recommender systems, robust question answering) on "deep pattern recognition".
- ozy 8y agoI believe we can make thinking machines. And "deep pattern matching" will be part of it. But it will need to fit in another architecture, most likely a lot like (deep) reinforcement learning. Probably with strong focus on modeling and predicting the environment. But also with some kind of goal modeling. See also for this not-magic: https://psyarxiv.com/387h9 https://psyarxiv.com/387h9
- ollin 8y agoBrains are not magic, and will be reproduced eventually, but DNNs are a fundamentally weaker architecture and won't be enough. Neural nets can solve some problems that brains can solve easily and lots of other ML methods couldn't solve, which is great. But the space of problems that brains can solve and neural nets can't is still rich, and will remain so until better methods are developed.