Deep Learning for Numerical Applications with SAS

Table of contents

1. Preface
2. About This Book
3. About The Author
4. Acknowledgments
5. Chapter 1: Introduction
   1. Deep Learning
   2. Is Deep Learning for You?
   3. It’s All about Performance
      1. Flynn’s Taxonomy
      2. Life after Flynn
   4. Organization of This Book
6. Chapter 2: Deep Learning
   1. Deep Learning
      1. Connectionism
      2. The Perceptron
      3. The First AI Winter
      4. The Experts to the Rescue
      5. The Second AI Winter
      6. The Deeps
      7. The Third AI Winter
      8. Some Supervision Required
   2. A Few Words about CAS
      1. Deployment Models
      2. CAS Sessions
      3. Caslibs
      4. Workers
      5. Action Sets and Actions
      6. Cleanup
   3. All about the Data
      1. The Men Body Mass Index Data Set
      2. The IRIS Data Set
   4. Logistic Regression
      1. Preamble
      2. Create the ANN
      3. Training
      4. Inference
   5. Conclusion
7. Chapter 3: Regressions
   1. A Brief History of Regressions
   2. All about the Data (Reprise)
      1. The CARS Data Set
   3. A Simple Regression
   4. The Universal Approximation Theorem
      1. Universal Approximation Framework
   5. Approximation of a Continuous Function
   6. Conclusions
8. Chapter 4: Many-Task Computing
   1. A Taxonomy for Parallel Programs
   2. Tasks Are the New Threads
      1. What Is a Task?
      2. Inputs and Outputs
      3. Immutable Inputs
      4. What Is a Job Flow?
   3. Examples of Job Flows
   4. Mutable Inputs
   5. Task Revisited
   6. Partitioning
   7. Federated Areas
   8. Persistent Area
   9. Caveats and Pitfalls
      1. Not Declaring Your Inputs
      2. Not Treating Your Immutable Inputs as Immutable
      3. Not Declaring Your Outputs
   10. Performance of Grid Scheduling
   11. Data-Object Pooling
   12. Portable Learning
   13. Conclusion
9. Chapter 5: Monte Carlo Simulations
   1. Monte Carlo or Las Vegas?
   2. Random Walk
   3. Multi-threaded Random Walk
      1. SAS Studio
      2. Live ETL
      3. A Parallel Program
      4. A Parallel Program with Partitions
      5. Many Cores
   4. Conclusion
10. Chapter 6: GPU
    1. History of GPUs
       1. The Golden Age of the Multicore
       2. The Golden Age of the Graphics Card
       3. The Golden Age of the GPU
    2. The CUDA Programming Model
    3. Hello π
       1. The CUDA Toolkit
       2. Buffon Revisited
    4. Generating Random Walk Data with CUDA
    5. Putting It All Together
    6. Conclusion
11. Chapter 7: Monte Carlo Simulations with Deep Learning
    1. Generating Data
       1. Training Data
       2. Testing Data
    2. Training the Network
    3. Inference Using the Network
    4. Performance Summary
    5. Other Examples
       1. Pricing of American Options
       2. Pricing of Variable Annuities Contracts
    6. Conclusion
12. Chapter 8: Deep Learning for Numerical Applications in the Enterprise
    1. Enterprise Applications
    2. A Task
       1. Data
       2. Task Implementation
    3. A Simple Flow
    4. A Training Flow Task
    5. An Inference Flow
    6. Documentation
    7. Heterogeneous Architectures
    8. Collaboration with Federated Areas
    9. Deploying DL with Federated Areas
    10. Conclusions
13. Chapter 9: Conclusions
    1. Data-Driven Programming
    2. The Quest for Speed
       1. From Tasks to GPUs
       2. Training and Inference
       3. FPGA
       4. Hybrid Architectures
14. Appendix A: Development Environment Setup
    1. LINUX
    2. Windows
15. References
16. Index