Group items matching

in title, tags, annotations or url

Video: https://www.youtube.com/watch?v=ElnuwpQEqTA

Seems to be a fun playground for Spiking Neural Networks right (from my newbie PoV)?

Might be of interest for torchquad and other projects

Let's keep a eye on this... Animals use odor cues to navigate through their environments, helping them locate targets and assess danger. Much of how animal brains organize, read out, and respond to odor stimuli across spatial and temporal scales is not well understood. To tackle these questions, Odor2Action uses a highly interdisciplinary team science approach. Our work uses fruit fly, honeybee, and mouse models to determine how neural representations of odor are generated, reformatted, and translated to generate useful behaviors that guide how animals interact with their environment.

reminds me of the methan smelling source finding study we did ...

The generated inputs are quite strange, I wonder where else something like this occurs.

Fascinating insight into the question of how networks learn.

Random branches are collected, scanned to 3D, and connected with servos. Then a neural network is trained to control this "robot".

If it can be done for obesity, I guess noise is also an option right? :)

love it

Just read some of the samples of text generated with their neural networks, insane.

It's really lucky that it was OpenAI who made that development and Elon Musk is so worried about AI. This way at least they try to assess the whole spectrum of abilities and applications of this model before releasing the full research to the public.

They released a smaller model, I got it running on Sandy. It's fairly straight forward: https://github.com/openai/gpt-2

Really nice visualizations for rectifier (ReLU) neural nets, illustrating the effects of skip-connections, depth, width, etc. on the loss function curvature.

Faced with a navigational challenge, neural networks spontaneously evolved units resembling the grid cells that help living animals find their way.

Dataset automatically constructed from public images uploaded by users on FB/Instagram with hashtags used as labels! They refer to this as 'weakly supervised learning'. Then neural net fine-tuned for ImageNet and achieved record 85.4% accuracy.

The race for the AI chips intensifies.

This has attracted a lot of attention in the DL community lately. Article: https://lilianweng.github.io/lil-log/2017/09/28/anatomize-deep-learning-with-information-theory.html Paper: https://arxiv.org/pdf/1703.00810.pdf

Neural Networks to analyse sleeplessness.

Genetic algorithms are a competitive alternative for training deep neural networks for reinforcement learning. For paper see: https://arxiv.org/pdf/1712.06567.pdf

Interesting pointers in this one! I would like to explore neuroevolution as well, although it seems extremely resource-demanding?

Not necessarily, I think it can be made to be much faster hybridizing it with backprop and Taylor maps. Its one ideas in the closet we still have not explored (Differential Intelligence: accelerating neuroevolution).

This has been around for over a year. The current trend in deep learning is "deeper is better". But a consequence of this is that for a given network depth, we can only feasibly evaluate a tiny fraction of the "search space" of NN architectures. The current approach to choosing a network architecture is to iteratively add more layers/units and keeping the architecture which gives an increase in the accuracy on some held-out data set i.e. we have the following information: {NN, accuracy}. Clearly, this process can be automated by using the accuracy as a 'signal' to a learning algorithm. The novelty in this work is they use reinforcement learning with a recurrent neural network controller which is trained by a policy gradient - a gradient-based method. Previously, evolutionary algorithms would typically be used. In summary, yes, the results are impressive - BUT this was only possible because they had access to Google's resources. An evolutionary approach would probably end up with the same architecture - it would just take longer. This is part of a broader research area in deep learning called 'meta-learning' which seeks to automate all aspects of neural network training.

Btw that techxplore article was cringing to read - if interested read this article instead: https://research.googleblog.com/2017/05/using-machine-learning-to-explore.html

"Hinton's new approach, known as capsule networks, is a twist on neural networks intended to make machines better able to understand the world through images or video. In one of the papers posted last week, Hinton's capsule networks matched the accuracy of the best previous techniques on a standard test of how well software can learn to recognize handwritten digits." Links to papers: https://arxiv.org/abs/1710.09829 https://openreview.net/forum?id=HJWLfGWRb&noteId=HJWLfGWRb

impressive!

seems a very impressive guy :"Hinton formed his intuition that vision systems need such an inbuilt sense of geometry in 1979, when he was trying to figure out how humans use mental imagery. He first laid out a preliminary design for capsule networks in 2011. The fuller picture released last week was long anticipated by researchers in the field. "Everyone has been waiting for it and looking for the next great leap from Geoff," says Kyunghyun Cho, a professor"