Autoplay
Autocomplete
Previous Lesson
Complete and Continue
314. Neural Network Optimization
Let's Get Started
In Our Last Episode...
Cottonwood Code Map
1. Single Hyperparameter Optimization
1.1 What is hyperparameter optimization? (3:50)
1.2 Code for testing repeatability (4:24)
1.3 Repeatability results (5:33)
1.4 Optimize a single hyperparameter (14:05)
2. Multi-Hyperparameter Optimization
2.1 Optimize two hyperparameters at once (12:09)
2.2 Refactor to separate out the optimization code (5:46)
2.3 Refactor to make optimization code object oriented (3:54)
2.4 Random search (6:54)
3. Evolutionary Powell's Method: An Experimental Optimizer
3.1 How Evolutionary Powell's method works
3.2 Evolutionary Powell's method in code (14:45)
3.3 Evolutionary Powell's' method results (9:26)
4. Code Optimization
4.1 Profiling (3:27)
4.2 Elimination and modification (5:35)
4.3 Cacheing and acceleration (5:27)
4.4 Pre-loading (4:42)
4.5 Parallelization (8:54)
5. Optimize the compressor
5.1 Hyperparameter optimization criterion: Compressed file size (6:31)
5.2 Neural architecture search setup (3:55)
5.3 Neural architecture search results (7:40)
5.4 Learning rate and initializer optimization (8:53)
6. Create an image compressor tool
6.1 API design for complete beginners (4:54)
6.2 API design for the image compressor (3:55)
6.3 API functions and arguments for the image compressor (9:18)
6.4 Image compressor tool code walkthrough, Part I (10:12)
6.5 Image compressor tool code walkthrough, Part II (5:44)
6.6 Test run the compressor and decompressor (4:50)
6.7 Inspect the results (5:16)
6.8 Improve the image quality (5:41)
6.9 Publish the Martian images case study (9:25)
6.3 API functions and arguments for the image compressor
Lesson content locked
If you're already enrolled,
you'll need to login
.
Enroll in Course to Unlock