PDF Publication Title:
Text from PDF Page: 025
2.6 policy gradients 17 (2) often the policy prematurely converges to a nearly-deterministic policy with a subop- timal behavior. Simple methods to prevent this issue, such as adding an entropy bonus, usually fail. The next two chapters in this thesis improve on the vanilla policy gradient method in two orthogonal ways, enabling us to obtain strong empirical results. Chapter 3 shows that instead of stepping in the gradient direction, we should move in the natural gradient direction, and that there is an effective way to choose stepsizes for reliable monotonic im- provement. Chapter 4 provides much more detailed analysis of discounts, and Chapter 5 also revisits some of the variance reduction ideas we have just described, but in a more general setting. Concurrently with this thesis work, Mnih et al. [Mni+16] have shown that it is in fact possible to obtain state-of-the-art performance on various large-scale control tasks with the vanilla policy gradient method, however, the number of samples used for learning is extremely large.PDF Image | OPTIMIZING EXPECTATIONS: FROM DEEP REINFORCEMENT LEARNING TO STOCHASTIC COMPUTATION GRAPHS
PDF Search Title:
OPTIMIZING EXPECTATIONS: FROM DEEP REINFORCEMENT LEARNING TO STOCHASTIC COMPUTATION GRAPHSOriginal File Name Searched:
thesis-optimizing-deep-learning.pdfDIY PDF Search: Google It | Yahoo | Bing
Cruise Ship Reviews | Luxury Resort | Jet | Yacht | and Travel Tech More Info
Cruising Review Topics and Articles More Info
Software based on Filemaker for the travel industry More Info
The Burgenstock Resort: Reviews on CruisingReview website... More Info
Resort Reviews: World Class resorts... More Info
The Riffelalp Resort: Reviews on CruisingReview website... More Info
| CONTACT TEL: 608-238-6001 Email: greg@cruisingreview.com | RSS | AMP |