Every other day we hear about new ways to put deep learning to good …
Every other day we hear about new ways to put deep learning to good use: improved medical imaging, accurate credit card fraud detection, long range weather forecasting, and more. PyTorch puts these superpowers in your hands, providing a comfortable Python experience that gets you started quickly and then grows with you as you—and your deep learning skills—become more sophisticated. Deep Learning with PyTorch will make that journey engaging and fun.
Although many deep learning tools use Python, the PyTorch library is truly Pythonic. Instantly familiar to anyone who knows PyData tools like NumPy and scikit-learn, PyTorch simplifies deep learning without sacrificing advanced features. It's excellent for building quick models, and it scales smoothly from laptop to enterprise. Because companies like Apple, Facebook, and JPMorgan Chase rely on PyTorch, it's a great skill to have as you expand your career options. It's easy to get started with PyTorch. It minimizes cognitive overhead without sacrificing the access to advanced features, meaning you can focus on what matters the most - building and training the latest and greatest deep learning models and contribute to making a dent in the world. PyTorch is also a snap to scale and extend, and it partners well with other Python tooling. PyTorch has been adopted by hundreds of deep learning practitioners and several first-class players like FAIR, OpenAI, FastAI and Purdue.
Deep Learning with PyTorch teaches you to create neural networks and deep learning systems with PyTorch. This practical book quickly gets you to work building a real-world example from scratch: a tumor image classifier. Along the way, it covers best practices for the entire DL pipeline, including the PyTorch Tensor API, loading data in Python, monitoring training, and visualizing results. After covering the basics, the book will take you on a journey through larger projects. The centerpiece of the book is a neural network designed for cancer detection. You'll discover ways for training networks with limited inputs and start processing data to get some results. You'll sift through the unreliable initial results and focus on how to diagnose and fix the problems in your neural network. Finally, you'll look at ways to improve your results by training with augmented data, make improvements to the model architecture, and perform other fine tuning.
For Python programmers with an interest in machine learning.
Eli Stevens had roles from software engineer to CTO, and is currently working on machine learning in the self-driving-car industry. Luca Antiga is cofounder of an AI engineering company and an AI tech startup, as well as a former PyTorch contributor. Thomas Viehmann is a PyTorch core developer and machine learning trainer and consultant.
Deep learning divided into digestible chunks with code samples that build up logically.Part 1. Core PyTorch
Chapter 1. Introducing deep learning and the PyTorch Library
Chapter 1. Why PyTorch?
Chapter 1. An overview of how PyTorch supports deep learning projects
Chapter 1. Hardware and software requirements
Chapter 2. Pretrained networks
Chapter 2. Obtaining a pretrained network for image recognition
Chapter 2. Ready, set, almost run
Chapter 2. A pretrained model that fakes it until it makes it
Chapter 2. A network that turns horses into zebras
Chapter 2. A pretrained network that describes scenes
Chapter 3. It starts with a tensor
Chapter 3. Indexing tensors
Chapter 3. The tensor API
Chapter 3. Tensor metadata: Size, offset, and stride
Chapter 3. NumPy interoperability
Chapter 4. Real-world data representation using tensors
Chapter 4. 3D images: Volumetric data
Chapter 4. Representing scores
Chapter 4. Working with time series
Chapter 4. Ready for training
Chapter 4. One-hot encoding whole words
Chapter 4. Text embeddings as a blueprint
Chapter 5. The mechanics of learning
Chapter 5. Gathering some data
Chapter 5. Down along the gradient
Chapter 5. Normalizing inputs
Chapter 5. Optimizers a la carte
Chapter 5. Generalizing to the validation set
Chapter 6. Using a neural network to fit the data
Chapter 6. More activation functions
Chapter 6. The PyTorch nn module
Chapter 6. Finally a neural network
Chapter 7. Telling birds from airplanes: Learning from images
Chapter 7. Distinguishing birds from airplanes
Chapter 7. Representing the output as probabilities
Chapter 7. Training the classifier
Chapter 7. The limits of going fully connected
Chapter 8. Using convolutions to generalize
Chapter 8. Convolutions in action
Chapter 8. Looking further with depth and pooling
Chapter 8. Subclassing nn.Module
Chapter 8. Training our convnet
Chapter 8. Helping our model to converge and generalize: Regularization
Chapter 8. Going deeper to learn more complex structures: Depth
Chapter 8. Comparing the designs from this section
Part 2. Learning from images in the real world: Early detection of lung cancer
Chapter 9. Using PyTorch to fight cancer
Chapter 9. What is a CT scan, exactly?
Chapter 9. In more detail, we will do the following
Chapter 9. Why canât we just throw data at a neural network until it works?
Chapter 9. What is a nodule?
Chapter 10. Combining data sources into a unified dataset
Chapter 10. Training and validation sets
Chapter 10. Loading individual CT scans
Chapter 10. Locating a nodule using the patient coordinate system
Chapter 10. A straightforward dataset implementation
Chapter 10. Constructing our dataset in LunaDataset.init
Chapter 11. Training a classification model to detect suspected tumors
Chapter 11. Pretraining setup and initialization
Chapter 11. Our first-pass neural network design
Chapter 11. The full model
Chapter 11. Outputting performance metrics
Chapter 11. Needed data for training
Chapter 11. Running TensorBoard
Chapter 11. Why isnât the model learning to detect nodules?
Chapter 12. Improving training with metrics and augmentation
Chapter 12. Graphing the positives and negatives
Chapter 12. Our ultimate performance metric: The F1 score
Chapter 12. What does an ideal dataset look like?
Chapter 12. Samplers can reshape datasets
Chapter 12. Revisiting the problem of overfitting
Chapter 12. Seeing the improvement from data augmentation
Chapter 13. Using segmentation to find suspected nodules
Chapter 13. Semantic segmentation: Per-pixel classification
Chapter 13. Updating the model for segmentation
Chapter 13. Updating the dataset for segmentation
Chapter 13. Building the ground truth data
Chapter 13. Implementing Luna2dSegmentationDataset
Chapter 13. Designing our training and validation data
Chapter 13. Updating the training script for segmentation
Chapter 13. Getting images into TensorBoard
Chapter 13. Results
Chapter 14. End-to-end nodule analysis, and where to go next
Chapter 14. Bridging CT segmentation and nodule candidate classification
Chapter 14. Did we find a nodule? Classification to reduce false positives
Chapter 14. Quantitative validation
Chapter 14. Reusing preexisting weights: Fine-tuning
Chapter 14. What we see when we diagnose
Chapter 14. Beyond a single best model: Ensembling
Chapter 14. Conclusion
Part 3. Deployment
Chapter 15. Deploying to production
Chapter 15. Request batching
Chapter 15. Exporting models
Chapter 15. Interacting with the PyTorch JIT
Chapter 15. TorchScript
Chapter 15. LibTorch: PyTorch in C++
Chapter 15. C++ from the start: The C++ API
Chapter 15. Going mobile
Chapter 15. Improving efficiency: Model design and quantization