Powered by TensorFlow: utilizing deep learning to better predict extreme weather 

I was working on predicting typhoon intensity in Japan using deep learning as well!

0:40 When people tell you that spaces are better than tabs, show them this frame. Now I will have nightmares

Just stumbled upon your channel. Thank y'all for the amazing work you do.

Harnessing the exceptional performance of the TensorFlow back end as well as the improved data feeding mechanisms through the tf data API, we achieve up to 40 TeraFLOP/s performance on a single nvidia v100 GPU. By applying optimizations to the distributed training enabler framework Horovod, the training algorithm as well as to the data feeding pipeline asynchronously executed on the IBM Power9 host processors, we could scale the distributed training efficiently to the full Summit AI system. Code delivered a sustained throughput of 325.8 PF/s and a parallel efficiency of 90.7% in single precision on 4560 Summit nodes, i.e. 27360 V100 GPUs. By taking advantage of the FP16 Tensor Cores, a half-precision version of the DeepLabv3+ network achieves a peak and sustained throughput of 1.13 EF/s and 999.0 PF/s respectively. Getting Access to Summit AI super computer : www.linkedin.com/pulse/president-trump-signs-development-ai-ganesan-narayanasamy

*Science For The Win!*

Who predicts better? IBM Watson (weather channel) or tensor flow?!?!?! Or are they different but with the same statistical out come?

Can the system predict earthquake?

it is still unpredictiable, we can built space craft and send rockets to moon even mars but if there is an unpredicted wind blown, then all those sophisticated vechiles cant launch :D