18 Issues in Current Deep Reinforcement Learning from ZhiHu

  • 时间:
  • 浏览:2
  • 来源:uu快3教程_uu快3代理_手机版

比较新的解法一个多多:

Ravi, S. and Larochelle, H. (2017). Optimization as a model for few-shot learning. In the International Conference on Learning Representations (ICLR).

Lillicrap, T. P., Hunt, J. J., Pritzel, A., Heess, N., Erez, T., & Tassa, Y., et al. (2015). Continuous control with deep reinforcement learning. Computer Science, 8(6), A187.

Sutton, R. S., Modayil, J., Delp, M., Degris, T., Pilarski, P. M., White, A., and Precup, D. (2011). Horde: A scalable real-time architecture for learning knowledge from unsupervised sensorimotor interaction, , proc. of 10th. In International Conference on Autonomous Agents and Multiagent Systems (AAMAS).

极其优秀的工作:unsupervised reinforcement and auxiliary learning 【Jaderberg et al 2017】

Tessler, C., Givony, S., Zahavy, T., Mankowitz, D. J., and Mannor, S. (2017). A deep hierarchical approach to lifelong learning in minecraft. In the AAAI Conference on Artificial Intelligence (AAAI).

Audiffren, J., Valko, M., Lazaric, A., and Ghavamzadeh, M. (2015). Maximum entropy semisupervised inverse reinforcement learning. In the International Joint Conference on Artificial Intelligence (IJCAI).

分水岭论文Deep Q-learning Network【Mnih et al 2013】中提到:并不一定当你们 的结果看上去很好,否则 那末任何理论依据(原文很狡猾的反过来说一遍)。

model-free planning

van Hasselt, H. (2010). Double Q-learning. Advances in Neural Information Processing Systems 23:, Conference on Neural Information Processing Systems 2010.

Tips:阅读此文请掌握DQN、Double DQN、Prioritized Experience Replay这个一个多多背景。

现有解法基本上围绕模仿学习

目前解法一个多多流派,一图胜千言:

deep exploration via bootstrapped DQN 【Osband et al 2016)】

异步算法A3C 【Mnih 2016】

under-appreciated reward exploration 【Nachum et al 2017)】

美中不够,TD Learning中很容易经常出先Over-Estimate(高估)什么的问题,具体愿因如下:

control, finding optimal policy

Mnih, V., Badia, A. P., Mirza, M., Graves, A., Harley, T., Lillicrap, T. P., Silver, D., and Kavukcuoglu, K. (2016). Asynchronous methods for deep reinforcement learning. In the International Conference on Machine Learning (ICML)

learn knowledge from different domains

Distributed Proximal Policy Optimization 【Heess 2017】

Bellemare, M. G., Danihelka, I., Dabney, W., Mohamed, S.,Lakshminarayanan, B., Hoyer, S., and Munos, R. (2017). The Cramer Distance as a Solution to Biased Wasserstein Gradients. ArXiv e-prints.

经验回放下的actor-critic 【Wang et al 2017b】

Zhu, X. and Goldberg, A. B. (309). Introduction to semi-supervised learning. Morgan & Claypool

现有解法围绕着无监督学习开展

吴恩达的逆强化学习【Ng and Russell 30)】

learn to navigate with unsupervised auxiliary learning 【Mirowski et al 2017】

Pan, S. J. and Yang, Q. (2010). A survey on transfer learning. IEEE Transactions on Knowledge and Data Engineering, 22(10):1345 – 1359.

Koch, G., Zemel, R., and Salakhutdinov, R. (2015). Siamese neural networks for one-shot image recognition. In the International Conference on Machine Learning (ICML).

He, F. S., Liu, Y., Schwing, A. G., and Peng, J. (2017a). Learning to play in a day: Faster deep reinforcement learning by optimality tightening. In the International Conference on Learning Representations (ICLR)

早在1997年Tsitsiklis就证明了肯能Function Approximator采用了神经网络这个非线性的黑箱,那末其收敛性和稳定性是无法保证的。

Schaul, T., Quan, J., Antonoglou, I., and Silver, D. (2016). Prioritized experience replay. In the International Conference on Learning Representations (ICLR).

万变不离其宗,Temporal Difference依据仍然是策略评估的核心哲学【Sutton 1988】。TD的拓展版本和她本身一样鼎鼎大名——1992年的Q-learning与2015年的DQN。

Sutton, R. S., McAllester, D., Singh, S., and Mansour, Y. (30). Policy gradient methods for reinforcement learning with function approximation. In the Annual Conference on Neural Information Processing Systems

(NIPS)
.

Wang, J. X., Kurth-Nelson, Z., Tirumala, D., Soyer, H., Leibo, J. Z., Munos, R., Blundell, C., Kumaran, D., and Botvinick, M. (2016a). Learning to reinforcement learn. arXiv:1611.05763v1.

现有解法是Guided Policy Search 【Levine et al 2016a】

Silver, D., Lever, G., Heess, N., Degris, T., Wierstra, D., & Riedmiller, M. (2014). Deterministic policy gradient algorithms. International Conference on International Conference on Machine Learning (pp.387-395). JMLR.org.

Mirowski, P., Pascanu, R., Viola, F., Soyer, H., Ballard, A., Banino, A., Denil, M., Goroshin, R., Sifre, L., Kavukcuoglu, K., Kumaran, D., and Hadsell, R. (2017). Learning to navigate in complex environments. In the International Conference on Learning Representations (ICLR).

Levine, S., Finn, C., Darrell, T., and Abbeel, P. (2016a). End-to-end training of deep visuomotor policies. The Journal of Machine Learning Research, 17:1–40.

benefit from non-reward training signals in environments

Silver, D., Huang, A., Maddison, C. J., Guez, A., Sifre, L., Van Den Driessche, G., Schrittwieser, J., Antonoglou, I., Panneershelvam, V., Lanctot, M., et al. (2016a). Mastering the game of go with deep neural networks and tree search. Nature, 529(7587):484–489.

Vezhnevets, A. S., Mnih, V., Agapiou, J., Osindero, S., Graves, A., Vinyals, O., and Kavukcuoglu, K. (2016). Strategic attentive writer for learning macro-actions. In the Annual Conference on Neural Information Processing Systems (NIPS).

learn from demonstration 【Hester et al 2017】

integrate temporal abstraction with intrinsic motivation 【Kulkarni et al 2016】

lifelong learning with hierarchical RL 【Tessler et al 2017】

Florensa, C., Duan, Y., and Abbeel, P. (2017). Stochastic neural networks for hierarchical reinforcement learning. In the International Conference on Learning Representations (ICLR)

Mahmood, A. R., van Hasselt, H., and Sutton, R. S. (2014). Weighted importance sampling for off-policy learning with linear function approximation. In the Annual Conference on Neural Information Processing Systems (NIPS).

Mnih, V., Kavukcuoglu, K., Silver, D., Rusu, A. A., Veness, J., Bellemare, M. G., Graves, A.,

Riedmiller, M., Fidjeland, A. K., Ostrovski, G., Petersen, S., Beattie, C., Sadik, A., Antonoglou, I., King, H., Kumaran, D., Wierstra, D., Legg, S., and Hassabis, D. (2015).

Human-level control through deep reinforcement learning. Nature, 518(7540):529–533.

Tamar, A., Wu, Y., Thomas, G., Levine, S., and Abbeel, P. (2016). Value iteration networks. In the Annual Conference on Neural Information Processing Systems (NIPS).

Kulkarni, T. D., Narasimhan, K. R., Saeedi, A., and Tenenbaum, J. B. (2016). Hierarchical deep reinforcement learning: Integrating temporal abstraction and intrinsic motivation. In the Annual Conference on Neural Information Processing Systems (NIPS)

现有解法:多层强化学习 【Barto and Mahadevan 303】

现有解法:

Lin, L. J. (1993). Reinforcement learning for robots using neural networks.

Stadie, B. C., Abbeel, P., and Sutskever, I. (2017).Third person imitation learning. In the International Conference on Learning Representations (ICLR).

focus on salient parts

Sutton, R. S. (1988). Learning to predict by the methods of temporal differences. Machine Learning,3(1):9–44.

旷世猛将van Hasselt先生很喜欢正确处理Over-Estimate什么的问题,他先搞出一个多多Double Q-learning【van Hasselt 2010】大闹NIPS,六年后搞出角度学习版本的Double DQN【van Hasselt 2016a】!

strategic attentive writer to learn macro-actions 【Vezhnevets et al 2016】

Finn, C., Christiano, P., Abbeel, P., and Levine, S. (2016a). A connection between GANs, inverse reinforcement learning, and energy-based models. In NIPS 2016 Workshop

on Adversarial Training.

Q-Prop, policy gradient with off-policy critic 【Gu et al 2017】

新的架构有【Kaiser et al 2017a、Silver et al 2016b、Tamar et al 2016、Vaswani et al 2017、Wang et al 2016c】

Nachum, O., Norouzi, M., Xu, K., and Schuurmans, D. (2017). Bridging the Gap Between Value and Policy Based Reinforcement Learning. ArXive-prints.

Emphatic-TD 【Sutton 2016】

现有解法有:

Mnih, Volodymyr, Kavukcuoglu, Koray, Silver, David, Graves, Alex, Antonoglou, Ioannis, Wier- stra, Daan, and Riedmiller, Martin. Playing atari with deep reinforcement learning. arXiv preprint arXiv:1312.5302, 2013.

Anschel, O., Baram, N., and Shimkin, N. (2017). Averaged-DQN: Variance reduction and stabilization for deep reinforcement learning. In the International Conference on Machine Learning (ICML).

GTD 【Sutton 309a、Sutton 309b、Mahmood 2014】

Schulman, J., Levine, S., Moritz, P., Jordan, M. I., and Abbeel, P. (2015). Trust region policy optimization. In the International Conference on Machine Learning (ICML).

data/sample efficiency

现有解法删改围绕迁移学习走 【Taylor and Stone, 309、Pan and Yang 2010、Weiss et al 2016】,learn invariant features to transfer skills 【Gupta et al 2017】

Gupta, A., Devin, C., Liu, Y., Abbeel, P., and Levine, S. (2017). Learning invariant feature spaces to transfer skills with reinforcement learning. In the International Conference on Learning Representations (ICLR).

Nachum, O., Norouzi, M., and Schuurmans, D. (2017). Improving policy gradient by exploring under-appreciated rewards. In the International Conference on Learning Representations (ICLR).

train dialogue policy jointly with reward model 【Su et al 2016b】

O'Donoghue, B., Munos, R., Kavukcuoglu, K., and Mnih, V. (2017). PGQ: Combining policy gradient and q-learning. In the International Conference on Learning Representations (ICLR).

现有解法有:

Horde 【Sutton et al 2011】

Lake, B. M., Salakhutdinov, R., and Tenenbaum, J. B. (2015). Human-level concept learning through probabilistic program induction. Science, 330(6266):1332–1338.

learning to learn, 【Duan et al 2017、Wang et al 2016a、Lake et al 2015】

Schulman, J., Abbeel, P., and Chen, X. (2017). Equivalence Between Policy Gradients and Soft Q-Learning. ArXiv e-prints.

Graves, A., Wayne, G., Reynolds, M., Harley, T., Danihelka, I., Grabska-Barwinska, A., Col- ´ menarejo, S. G., Grefenstette, E., Ramalho, T., Agapiou, J., nech Badia, A. P., Hermann, K. M., Zwols, Y., Ostrovski, G., Cain, A., King, H., Summerfield, C., Blunsom, P., Kavukcuoglu, K., and Hassabis, D. (2016). Hybrid computing using a neural network with dynamic external memory. Nature, 538:471–476

Munos, R., Stepleton, T., Harutyunyan, A., and Bellemare, M. G.(2016). Safe and efficient offpolicy reinforcement learning. In the Annual Conference on Neural Information Processing Systems (NIPS).

Gruslys, A., Gheshlaghi Azar, M., Bellemare, M. G., and Munos, R. (2017). The Reactor: A Sample-Efficient Actor-Critic Architecture. ArXiv e-prints

Watkins, C. J. C. H. and Dayan, P. (1992). Q-learning. Machine Learning, 8:279–292

Xu, K., Ba, J. L., Kiros, R., Cho, K., Courville, A.,Salakhutdinov, R., Zemel, R. S., and Bengio,Y. (2015). Show, attend and tell: Neural image caption generation with visual attention. In the International Conference on Machine Learning (ICML).

Sutton, R. S., Szepesvari, C., and Maei, H. R. (309b). A convergent O( ´ n) algorithm for off-policy temporal-difference learning with linear function approximation. In the Annual Conference on Neural Information Processing Systems (NIPS).

Heess, N., TB, D., Sriram, S., Lemmon, J., Merel, J., Wayne, G., Tassa, Y., Erez, T., Wang, Z., Eslami, A., Riedmiller, M., and Silver, D. (2017). Emergence of Locomotion Behaviours in Rich Environments. ArXiv e-prints

角度强化学习的什么的问题在哪里?未来如保会会走?哪些地方方面都都可以突破?

下面是CV和NLP方面的几个简介:物体检测 【Mnih 2014】、机器翻译 【Bahdanau 2015】、图像标注【Xu 2015】、用Attention代替CNN和RNN【Vaswani 2017】等等。

Vinyals, O., Blundell, C., Lillicrap, T., Kavukcuoglu, K., and Wierstra, D. (2016). Matching networks for one shot learning. In the Annual Conference on Neural Information Processing Systems (NIPS).

Sutton, R. S. (1990). Integrated architectures for learning, planning, and reacting based on approximating dynamic programming. In the International Conference on Machine Learning (ICML).

Jaderberg, M., Mnih, V., Czarnecki, W., Schaul, T., Leibo, J. Z., Silver, D., and Kavukcuoglu, K. (2017). Reinforcement learning with unsupervised auxiliary tasks. In the International Conference on Learning Representations (ICLR).

Sutton, R. S., Maei, H. R., Precup, D., Bhatnagar, S., Silver, D., Szepesvari, C., and Wiewiora, ´E. (309a). Fast gradient-descent methods for temporal-difference learning with linear function approximation. In the International Conference on Machine Learning (ICML).

TRPO(Trust Region Policy Optimization)【Schulman 2015】

Sutton老爷子教科书里的经典安利:Dyna-Q 【Sutton 1990】

Instability and Divergence when combining off-policy,function approximation,bootstrapping

train perception and control jointly end-to-end

Weiss, K., Khoshgoftaar, T. M., and Wang, D. (2016). A survey of transfer learning. Journal of Big Data, 3(9)

exploration-exploitation tradeoff

大名鼎鼎的GANs 【Goodfellow et al 2014】

Barto, A. G., Sutton, R. S., and Anderson, C. W. (1983). Neuronlike elements that can solve difficult learning control problems. IEEE Transactions on Systems, Man, and Cybernetics, 13:835–846

imitation learning with GANs 【Ho and Ermon 2016、Stadie et al 2017】 (其TensorFlow实现在imitation)

PGQ,policy gradient and Q-learning 【O'Donoghue et al 2017】

Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A., , and Bengio, Y. (2014). Generative adversarial nets. In the Annual

Conference on Neural Information Processing Systems (NIPS), page 2672?2630.

gigantic search space

Williams, R. J. (1992). Simple statistical gradient-following algorithms for connectionist reinforcement learning. Machine Learning, 8(3):229–256.

Kaiser, Ł., Nachum, O., Roy, A., and Bengio, S. (2017b). Learning to Remember Rare Events. In the International Conference on Learning Representations (ICLR).

Kaiser, L., Gomez, A. N., Shazeer, N., Vaswani, A., Parmar, N., Jones, L., and Uszkoreit, J. (2017a). One Model To Learn Them All. ArXiv e-prints.

Duel DQN【Wang 2016c】(ICML2016最佳论文)

DQN的改良主要依靠一个多多Trick:

下面几篇论文不是DQN相关话题的:

Li, K. and Malik, J. (2017). Learning to optimize. In the International Conference on Learning Representations (ICLR).

最出名的解法是在Nature上大秀一把的Differentiable Neural Computer【Graves et al 2016】

learn with MDPs both with and without reward functions 【Finn et al 2017)】

learn, plan, and represent knowledge with spatio-temporal abstraction at multiple levels

Chebotar, Y., Hausman, K., Zhang, M., Sukhatme, G., Schaal, S., and Levine, S. (2017). Combining model-based and model-free updates for trajectory-centric reinforcement learning. In the International Conference on Machine Learning (ICML)

return-based off-policy control, Retrace 【Munos et al 2016】, Reactor 【Gruslyset al 2017】

下面经常出先DQN的范畴——

Zoph, B. and Le, Q. V. (2017). Neural architecture search with reinforcement learning. In the International Conference on Learning Representations (ICLR)

data storage over long time, separating from computation

reward function not available

benefit from both labelled and unlabelled data

learn a flexible RNN model to handle a family of RL tasks 【Duan et al 2017、Wang et al 2016a】

Sutton, R. S., Mahmood, A. R., and White, M. (2016). An emphatic approach to the problem of off-policy temporal-difference learning. The Journal of Machine Learning Research, 17:1–29

Taylor, M. E. and Stone, P. (309). Transfer learning for reinforcement learning domains: A survey. Journal of Machine Learning Research, 10:1633–1685.

TODO list:文章内容还不够充实,否则 论文是全的。未来一段时间会把论文的链接找齐,下载好否则 打个包传到百度云上,预计一四天完成。(2017/12/19)

Baker, B., Gupta, O., Naik, N., and Raskar, R. (2017). Designing neural network architectures using reinforcement learning. In the International Conference on Learning Representations (ICLR).

one/few/zero-shot learning 【Duan et al 2017、Johnson et al 2016、 Kaiser et al 2017b、Koch et al 2015、Lake et al 2015、Li and Malik 2017、Ravi and Larochelle, 2017、Vinyals et al 2016)

adapt rapidly to new tasks

prediction, policy evaluation

unify count-based exploration and intrinsic motivation 【Bellemare et al 2017】

Osband, I., Blundell, C., Pritzel, A., and Roy, B. V. (2016). Deep exploration via bootstrapped DQN. In the Annual Conference on Neural Information Processing Systems (NIPS).

(neural networks architecture design )

model-free与model-based的结合使用【Chebotar et al 2017】

Silver, D., van Hasselt, H., Hessel, M., Schaul, T., Guez, A., Harley, T., Dulac-Arnold, G., Reichert, D., Rabinowitz, N., Barreto, A., and Degris, T. (2016b). The predictron: End-to-end learning and planning. In NIPS 2016 Deep Reinforcement Learning Workshop.

现有的网络架构搜索依据【Baker et al 2017、Zoph and Le 2017】,其中Zoph的工作分量非常重。

Johnson, M., Schuster, M., Le, Q. V., Krikun, M., Wu, Y., Chen, Z., Thorat, N., Viegas, F., Watten- ´berg, M., Corrado, G., Hughes, M., and Dean, J. (2016). Google’s Multilingual Neural Machine Translation System: Enabling Zero-Shot Translation. ArXive-prints.

Sutton, R. S. and Barto, A. G. (2017). Reinforcement Learning: An Introduction (2nd Edition, in preparation). MIT Press.

这四天我阅读了两篇篇猛文A Brief Survey of Deep Reinforcement Learning 和 Deep Reinforcement Learning: An Overview ,作者排山倒海的引用了30多篇文献,阐述强化学习未来的方向。原文归纳出角度强化学习中的常见科学什么的问题,并列出了目前解法与相关综述,我在这里做出架构设计 ,抽取了相关的论文。

stochastic neural networks for hierarchical RL 【Florensa et al 2017】

Bahdanau, D., Brakel, P., Xu, K., Goyal, A., Lowe, R., Pineau, J., Courville, A., and Bengio, Y. (2017). An actor-critic algorithm for sequence prediction. In the International

Conference on Learning Representations (ICLR).

现有解法基本上是learn to learn

这里精选18个关键什么的问题,涵盖空间搜索、探索利用、策略评估、内存使用、网络设计、反馈激励等等话题。本文精选了73篇论文(其中2017年论文有27篇,2016年论文有21篇)为了方便阅读,原标题插进文章最后,都都可以根据索引找到。

Hester, T. and Stone, P. (2017). Intrinsically motivated model learning for developing curious robots. Artificial Intelligence, 247:170–86.

variational information maximizing exploration 【Houthooft et al 2016】

Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, L., and Polosukhin, I. (2017). Attention Is All You Need. ArXiv e-prints.

现有解法依然是蒙特卡洛搜索,详情都都可以参考初代AlphaGo的实现【Silver et al 2016a】

Mnih, V., Heess, N., Graves, A., and Kavukcuoglu, K. (2014). Recurrent models of visual attention. In the Annual Conference on Neural Information Processing Systems

(NIPS)
.

van Hasselt, H., Guez, A., , and Silver, D. (2016a). Deep reinforcement learning with double Qlearning. In the AAAI Conference on Artificial Intelligence (AAAI).

Houthooft, R., Chen, X., Duan, Y., Schulman, J., Turck, F. D., and Abbeel, P. (2016). Vime: Variational information maximizing exploration. In the Annual Conference on Neural Information Processing Systems (NIPS).

Ng, A. and Russell, S. (30).Algorithms for inverse reinforcement learning. In the International Conference on Machine Learning (ICML).

Wang, Z., Schaul, T., Hessel, M., van Hasselt, H., Lanctot, M., and de Freitas, N. (2016c). Dueling network architectures for deep reinforcement learning. In the International

Conference on Machine Learning (ICML)
.

Barto, A. G. and Mahadevan, S. (303). Recent advances in hierarchical reinforcement learning. Discrete Event Dynamic Systems, 13(4):341–379.

Duan, Y., Andrychowicz, M., Stadie, B. C., Ho, J., Schneider, J.,Sutskever, I., Abbeel, P., and Zaremba, W. (2017). One-Shot Imitation Learning. ArXiv e-prints.

Policy gradient与Q-learning 的结合【O'Donoghue 2017、Nachum 2017、 Gu 2017、Schulman 2017】

否则 ,一个多多很好的思路是从计算机视觉与自然语言正确处理领域汲取灵感,这类下文中肯能提到的unsupervised auxiliary learning依据借鉴了RNN+LSTM中的极少量操作

learn with expert's trajectories and those may not from experts 【Audiffren et al 2015】

Ho, J. and Ermon, S. (2016). Generative adversarial imitation learning. In the Annual Conference on Neural Information Processing Systems (NIPS).

model-based learning

Gu, S., Lillicrap, T., Ghahramani, Z., Turner, R. E., and Levine, S. (2017). Q-Prop: Sampleefficient policy gradient with an off-policy critic. In the International

Conference on Learning Representations (ICLR).

现有解法删改围绕半监督学习 【Zhu and Goldberg 309】

Wang, S. I., Liang, P., and Manning, C. D. (2016b). Learning language games through interaction. In the Association for Computational Linguistics annual meeting (ACL)

Q-learning与Actor-Critic