2019 05 11 Chicago Codecamp - Deep Learning for everyone? Challenge Accepted!
Download as PPTX, PDF0 likes1,420 views
The document discusses deep learning and its applications, particularly focusing on image classification techniques using frameworks like TensorFlow and Keras. It outlines the process of building and training machine learning models, including data preparation, feature extraction, and model evaluation. The content emphasizes the importance of convolutional neural networks and their components, like activation functions and pooling layers, in enhancing image analysis capabilities.
2019 05 11 Chicago Codecamp - Deep Learning for everyone? Challenge Accepted!
- 1. Bruno Capuano Innovation Lead @Avanade @elbruno | http://elbruno.com Deep Learning for everyone? Challenge Accepted!
- 2. @elbruno IBM slaps patent on coffee-delivering drones (link)
- 3. @elbruno IBM slaps patent on coffee-delivering drones (link)
- 6. @elbruno Load Data Extract Features Model Consumption Train Model Evaluate Model Prepare Your Data Build & Train Run Machine Leaning workflow
- 7. @elbruno7
- 9. @elbruno9
- 12. @elbruno1
- 13. @elbruno1
- 15. @elbruno Boy / Girl Baby Teenager Old dude / Lady
- 16. @elbruno Load Data Extract Features Model Consumption Train Model Evaluate Model Prepare Your Data Build & Train Run Machine Leaning workflow
- 17. @elbruno Is this A or B? How much? How many? How is this organized? Regression ClusteringClassification Machine Learning Tasks
- 18. @elbruno1 Cognitive Services - Anomaly Detector
- 20. Deep Learning
- 21. @elbruno Machine Learning Seth Juarez - @sethjuarez - Microsoft
- 22. @elbruno Machine Learning Seth Juarez - @sethjuarez - Microsoft
- 23. @elbruno Deep Learning Seth Juarez - @sethjuarez - Microsoft
- 24. @elbruno Deep Learning Seth Juarez - @sethjuarez - Microsoft
- 25. @elbruno MNIST dataset • The MNIST database of handwritten digits, • Has a training set of 60,000 examples, and a test set of 10,000 examples. • It is a subset of a larger set available from NIST. • The digits have been size- normalized and centered in a fixed-size image.
- 26. @elbruno
- 27. @elbruno
- 28. @elbruno Artificial Intelligence: Image Analysis
- 29. @elbruno Why is this hard? You see this: But the camera sees this:
- 30. @elbruno Machine learning and feature representations Input Raw image Chihuahua “Non”-chihuahuas Learning algorithm pixel 1 pixel2 pixel 1 pixel 2
- 31. @elbruno Input What we want Learning algorithm pixel 1 pixel2 Feature representation ear eye E.g., Does it have 2 ears? 2 eyes? eye ear Raw image Features Chihuahua “Non”-chihuahuas
- 32. @elbruno A machine learning subfield of learning representations of data. Exceptional effective at learning patterns. Deep learning algorithms attempt to learn (multiple levels of) representation by using a hierarchy of multiple layers If you provide the system tons of information, it begins to understand it and respond in useful ways. Deep Learning (DL) https://www.xenonstack.com/blog/static/public/uploads/media/machine-learning-vs-deep-learning.png
- 33. @elbruno TensorFlow Playground Demo MakeMagicHappen(); https://www.avanade.com/AI
- 34. @elbruno
- 35. @elbruno Open source software library for numerical computation using data flow graphs Developed by Google Brain Team for machine learning and deep learning and made open-source TensorFlow provides an extensive suite of functions and classes that allow users to build various models from scratch What is TensorFlow? These slides are adapted from the following Stanford lectures: https://web.stanford.edu/class/cs20si/2017/lectures/slides_01.pdf https://cs224d.stanford.edu/lectures/CS224d-Lecture7.pdf
- 36. @elbruno import tensorflow as tf a = tf.add(2, 3) TF automatically names nodes if you do not x = 2 y = 3 print a >> Tensor("Add:0", shape=(), dtype=int32) Note: a is NOT 5 TensorFlow Graphs a
- 37. @elbruno Keras is a layer on top of TensorFlow that makes it much easier to create neural networks. It provides a higher level API for various machine learning routines. Unless you are performing research into entirely new structures of deep neural networks it is unlikely that you need to program TensorFlow directly. Keras is a separate install from TensorFlow. To install Keras, use pip install keras (after installing TensorFlow). TensorFlow and Keras
- 38. @elbruno Keras Sequential model is used to create a feed-forward network, by stacking layers (successive ‘add’ operations). Shape of the input layer is specified in the first hidden layer (or the output layer if network had no hidden layer). - Input 2D image is flattened to 1D vector. - Dropout (with the rate 0.2) is applied to the first hidden layer TensorFlow for Classification: MNIST
- 39. @elbrunohttp://deeplearning.stanford.edu/wiki/index.php/Feature_extraction_using_convolution Main CNN idea for text: Compute vectors for n-grams and group them afterwards A convolution layer is created by defining a kernel - which is fixed square window of weights, and scanning this across the whole input image. Input matrix Convolutional 3x3 filter Convolutional Neural Networks (CNNs)
- 40. @elbruno Pooling layer Pooling layers are a form of downsampling which usually follow convolution layers in the neural network. Applying pooling to a feature map transforms the map into a smaller representation, and it loses some of the exact positional information of the features. Therefore it makes our network more invariant to small transformations and distortions in the input image by asking whether a feature appears in a given region of an image (the pooling region) rather than at a specific location. The most common method of pooling is max-pooling, where the the maximum value of a given region in a feature map is taken. The example below shows max pooling being applied with a 2×2 pooling region. https://shafeentejani.github.io/assets/images/pooling.gif
- 41. @elbruno Activation: ReLU Takes a real-valued number and thresholds it at zero 𝑅 𝑛 → 𝑅+ 𝑛 Most Deep Networks use ReLU nowadays � Trains much faster • accelerates the convergence of SGD • due to linear, non-saturating form � Less expensive operations • compared to sigmoid/tanh (exponentials etc.) • implemented by simply thresholding a matrix at f 𝑥 = max(0, 𝑥) http://adilmoujahid.com/images/activation.png
- 42. @elbruno Deep Learning for image analysis
- 43. @elbruno Deep Neural Network: Cat vs Dog https://becominghuman.ai/building-an-image-classifier-using-deep-learning-in-python-totally-from-a-beginners-perspective-be8dbaf22dd8
- 44. @elbruno Deep Neural Network: Cat vs Dog https://becominghuman.ai/building-an-image-classifier-using-deep-learning-in-python-totally-from-a-beginners-perspective-be8dbaf22dd8
- 45. @elbruno Cat vs Dogs Demo MakeMagicHappen(); https://www.avanade.com/AI
- 46. @elbruno Bruno Capuano Innovation Lead @Avanade @elbruno | http://elbruno.com Q&A Thanks!