Deep Learning With Python & Tensorflow

on YouTube ↗49m

" summary

Shows how to build and train models with TensorFlow (Python), including a live MNIST demo that uses 28x28 = 784 input tensors, one hidden layer and 91% accuracy on MNIST with one hidden layer. Uses 100-sample minibatches from 55,000 images and gradient descent; reproduces a similar Theano result (~89%). Describes distributed training (50–500 nodes, 40x ImageNet speedups), cloud option that cuts an 8-hour job to 32 minutes on 20 nodes, and deploying with TensorFlow Serving and custom TPUs.

AI-written summary · about

" chapters (9)

1. 0:00Introduction and background
2. 3:04Deep learning fundamentals
3. 9:14Tensors and neural network mechanics
4. 15:01The evolution of machine learning
5. 23:40Introduction to TensorFlow
6. 26:43Building a model with TensorFlow
7. 35:30Training and performance testing
8. 43:16Distributed training and hardware
9. 47:21Conclusion and resources

" related talks

• 01h 06mDeep Learning with TensorFlowSeth Juarez · NDC Sydney 2018
• 01h 07mDeep Learning with TensorFlowSeth Juarez · NDC Oslo 2018
• 01h 29mHands-on introduction to Deep Learning with Keras and TensorflowRodrigo Agundez · PyData Amsterdam 2018
• 44mTensorflow, Tensorflow, What is it that you see?Ashic Mahtab · Oredev 2017
• 57mDeep Learning with C++Peter Goldsborough · Meeting C++ 2017
• 01h 02mNetworks are like onions: Practical Deep Learning with TensorFlowBarbara Fusinska · NDC London 2018
• 01h 05mDeep Learning with PyTorchSeth Juarez · NDC London 2019
• 34mML from a software engineer's perspectiveMarijn van Zelst · Codemotion Amsterdam 2017