This week’s show covers Google’s use of machine learning to cut datacenter power consumption, NVIDIA’s new ‘crazy, reckless’ GPU, and a new Layer Normalization technique that promises to reduce the training time for deep neural networks. Plus, a bunch more.
This week’s podcast is sponsored by Cloudera, organizers of the Wrangle Conference which is coming up in San Francisco on July 28th. Check out the event page for information on the great talks and speakers they’ve got planned, and if you decide to register, use the code “COMMUNITY” for 20% off!
Subscribe: iTunes / SoundCloud / Google Play / Stitcher / RSS
Here are the notes for this week’s podcast:
Google Drives Datacenter Efficiency with Deep Learning
- Google DeepMind
- machine-learning-applicationsfor-datacenter-optimization-finalv2.pdf
- Google Cuts Its Giant Electricity Bill With DeepMind-Powered AI – Bloomberg
Google’s Cloud Natural Language and Cloud Speech APIs in Public Beta
- <a href="https://cloudplatform.googleblog blood pressure medicine lisinopril.com/2016/07/the-latest-for-Cloud-customers-machine-learning-and-west-coast-expansion.html”>Google Cloud Platform Blog: Introducing Cloud Natural Language API, Speech API open beta and our West Coast region expansion
- Using the Cloud Natural Language API to analyze Harry Potter and The New York Times | Google Cloud Big Data and Machine Learning Blog | Google Cloud Platform
NVIDIA’s New “Crazy, Reckless” GPU
- The New NVIDIA TITAN X: The Ultimate. Period. | The Official NVIDIA Blog
- NVIDIA Announces “NVIDIA Titan X” Video Card: $1200, Available August 2nd
Sketchy Enterprise AI Adoption Numbers
- Outlook on Artificial Intelligence in the Enterprise 2016 [Research Report]
- AI adoption coming quickly to the enterprise sector
Layer Normalization for Faster RNN Training
- Layer Normalization Paper on arXiv [1607.06450v1.pdf]
- Batch Normalization Paper on arXiv [1502.03167v3.pdf]
- layer-norm/layers.py at master · ryankiros/layer-norm
- Keras GRU with Layer Normalization
TensorFlow With Latest RNN & NLP Papers
Projects
- Neural Network Learns to Generate Voice (RNN/LSTM) – YouTube
- How English sounds to non-English speakers – YouTube
- The Skynet Salesman | Stitch Fix Technology
- A Beginner’s Guide To Understanding Convolutional Neural Networks
- Approaching (Almost) Any Machine Learning Problem
Image: NVIDIA
Paddy
Hi Sam,
Really enjoying the podcast, keep up the good work.
As someone who is relatively new to the space (just finished the Titanic Kaggle), I really enjoyed the recommendation for Abhishek’s blog post Approaching (almost) any machine learning problem. Given that this post focuses on applying ML algorithms, is there anything that you know of that deals with the data munging / cleaning etc. that happens prior to this?
Thanks