2018-04-21

【入门必备】史上最全的深度学习资源汇总，速藏！

• 入门 | 通过 Q-learning 深入理解强化学习

• 学界 | Uber AI论文：利用反向传播训练可塑神经网络，生物启发的元学习范式

• 业界 | OpenAI提出新型元学习方法EPG，调整损失函数实现新任务上的快速训练

• Collabbing - Ideas - 强化学习路线

• 变形卷积核、可分离卷积？卷积神经网络中十大拍案叫绝的操作。

• 【入门必备】史上最全的深度学习资源汇总，速藏！

• 哈工大朱晓蕊教授：CCF-GAIR 2018 将会与你擦出怎样的火花

• 关于序列建模，是时候抛弃RNN和LSTM了 | 机器之心

• UC Berkeley提出新型「增强型随机搜索」算法，可提高连续控制问题中的样本效率

• Using Meta-Learning in Nonstationary and Competitive Environments with Pieter Abbeel et al | Packt Hub

• ICLR 2018最佳论文：基于梯度的元学习算法，可高效适应非平稳环境 | 机器之心

• Evolved Policy Gradients

• 入门 | 通过 Q-learning 深入理解强化学习

• 学界 | Uber AI论文：利用反向传播训练可塑神经网络，生物启发的元学习范式

• 业界 | OpenAI提出新型元学习方法EPG，调整损失函数实现新任务上的快速训练

• [1706.03762 Attention Is All You Need.pdf](file:///E:/%E6%90%9C%E7%8B%97%E9%AB%98%E9%80%9F%E4%B8%8B%E8%BD%BD2%202017-10-24/1706.03762%20Attention%20Is%20All%20You%20Need.pdf)

• 伯克利提出强化学习新方法，可让智能体同时学习多个解决方案 | 机器之心

• [1803.07055 Simple random search provides a competitive approach to reinforcement learning.pdf](file:///E:/%E6%90%9C%E7%8B%97%E9%AB%98%E9%80%9F%E4%B8%8B%E8%BD%BD2%202017-10-24/1803.07055%20Simple%20random%20search%20provides%20a%20competitive%20approach%20%20to%20reinforcement%20learning.pdf)

• [Continuous Adaptation via Meta-Learning in Nonstationary and Competitive Environments 2018 ICLR.pdf](file:///E:/%E6%90%9C%E7%8B%97%E9%AB%98%E9%80%9F%E4%B8%8B%E8%BD%BD2%202017-10-24/Continuous%20Adaptation%20via%20Meta-Learning%20in%20Nonstationary%20and%20Competitive%20Environments%202018%20ICLR.pdf)

• [Reinforcement Learning with Deep Energy-Based Policies](file:///E:/%E6%90%9C%E7%8B%97%E9%AB%98%E9%80%9F%E4%B8%8B%E8%BD%BD2%202017-10-24/1702.08165%20[[1702.08165]%20Reinforcement%20Learning%20with%20Deep%20Energy-Based%20Policies].pdf)

• 要合作，不要对抗！无需预训练超越经典算法，上交大提出合作训练式生成模型CoT

• [1804.03782 CoT Cooperative Training for Generative Modeling (2).pdf](file:///E:/%E6%90%9C%E7%8B%97%E9%AB%98%E9%80%9F%E4%B8%8B%E8%BD%BD2%202017-10-24/1804.03782%20CoT%20Cooperative%20Training%20for%20Generative%20Modeling%20(2).pdf)

• Sequence generative adversarial nets with policy gradient的相关微信公众号文章 – 搜狗微信搜索

• [1609.05473] SeqGAN: Sequence Generative Adversarial Nets with Policy Gradient

• How to Train your Generative Models? And why does Adversarial Training work so well?

• [SeqGAN Sequence Generative Adversarial Nets with Policy Gradient 1609.05473v5.pdf](file:///D:/[%E6%B7%B1%E5%BA%A6%E5%AD%A6%E4%B9%A0%E8%B5%84%E6%96%99][2016-2017]/SeqGAN%20Sequence%20Generative%20Adversarial%20Nets%20with%20Policy%20Gradient%201609.05473v5.pdf)

变形卷积核、可分离卷积？卷积神经网络中十大拍案叫绝的操作。