以下仅为个人评分。官方论文评分请参看：点击打开链接 包括2018年977篇研究论文
英文名：
Certifiable Distributional Robustness with Principled Adversarial Training 9.0
On the Convergence of Adam and Beyond 8.5
Emergence of grid-like representations by training recurrent neural networks to perform spatial localization 8.33333333333
Multi-Scale Dense Networks for Resource Efficient Image Classification 8.33333333333
Boosting Dilated Convolutional Networks with Mixed Tensor Decompositions 8.0
i-RevNet: Deep Invertible Networks 8.0
Learning to Represent Programs with Graphs 8.0
Continuous Adaptation via Meta-Learning in Nonstationary and Competitive Environments 8.0
Stabilizing Adversarial Nets with Prediction Methods 8.0
A DIRT-T Approach to Unsupervised Domain Adaptation 7.5
Alternating Multi-bit Quantization for Recurrent Neural Networks 7.5
Spherical CNNs | OpenReview 7.5
Deep Mean Field Games for Learning Optimal Behavior Policy of Large Populations 7.5
FusionNet: Fusing via Fully-aware Attention with Application to Machine Comprehension 7.5
Distributed Prioritized Experience Replay 7.33333333333
A PAC-Bayesian Approach to Spectrally-Normalized Margin Bounds for Neural Networks 7.33333333333
Unsupervised Machine Translation Using Monolingual Corpora Only 7.33333333333
CausalGAN: Learning Causal Implicit Generative Models with Adversarial Training 7.33333333333
GENERALIZING ACROSS DOMAINS VIA CROSS-GRADIENT TRAINING 7.33333333333
Neural Speed Reading via Skim-RNN 7.33333333333
Learning Differentially Private Recurrent Language Models 7.33333333333
Synthetic and Natural Noise Both Break Neural Machine Translation 7.33333333333
Learning One-hidden-layer Neural Networks with Landscape Design 7.33333333333
Neural Sketch Learning for Conditional Program Generation 7.33333333333
Modular Continual Learning in a Unified Visual Environment 7.33333333333
AmbientGAN : Generative models from lossy measurements 7.33333333333
Deep Autoencoding Gaussian Mixture Model for Unsupervised Anomaly Detection 7.33333333333
Decision-Based Adversarial Attacks: reliable attacks against Black-Box Machine Learning Models 7.33333333333
Skip Connections Eliminate Singularities 7.33333333333
Eigenoption Discovery through the Deep Successor Representation 7.33333333333
Learning how to explain neural networks: PatternNet and PatternAttribution 7.33333333333
Spectral Normalization for Generative Adversarial Networks 7.33333333333
Mastering the Dungeon: Grounded Language Learning by Mechanical Turker Descent 7.33333333333
PixelNN: Example-based Image Synthesis 7.0
Compressing Word Embeddings via Deep Compositional Code Learning 7.0
Variational Network Quantization 7.0
Learning from Between-class Examples for Deep Sound Recognition 7.0
Deep Learning for Physical Processes: Incorporating Prior Scientific Knowledge 7.0
On the importance of single directions for generalization 7.0
Training RNNs as Fast as CNNs 7.0
Variational image compression with a scale hyperprior 7.0
Deep Learning as a Mixed Convex-Combinatorial Optimization Problem 7.0
Spatially Transformed Adversarial Examples 7.0
Attacking Binarized Neural Networks 7.0
Generative Models of Visually Grounded Imagination 7.0
Variance Reduction for Policy Gradient Methods with Action-Dependent Baselines 7.0
Efficient Sparse-Winograd Convolutional Neural Networks 7.0
Training GANs with Optimism 7.0
Minimal-Entropy Correlation Alignment for Unsupervised Deep Domain Adaptation 7.0
N2N learning: Network to Network Compression via Policy Gradient Reinforcement Learning 7.0
Simulating Action Dynamics with Neural Process Networks 7.0
A Neural Representation of Sketch Drawings 7.0
Certified Defenses against Adversarial Examples 7.0
Lifelong Learning with Dynamically Expandable Networks 7.0
Backpropagation through the Void: Optimizing control variates for black-box gradient estimation 7.0
Neural Language Modeling by Jointly Learning Syntax and Lexicon 7.0
Learning Latent Permutations with Gumbel-Sinkhorn Networks 7.0
Zero-Shot Visual Imitation 7.0
Intrinsic Motivation and Automatic Curricula via Asymmetric Self-Play 7.0
Regularizing and Optimizing LSTM Language Models 7.0
Active Neural Localization 7.0
SAMPLE-EFFICIENT POLICY OPTIMIZATION WITH STEIN CONTROL VARIATE 7.0
Generalizing Hamiltonian Monte Carlo with Neural Networks 7.0
Analyzing the Role of Temporal Differencing in Deep Reinforcement Learning 7.0
A Scalable Laplace Approximation for Neural Networks 7.0
Hyperparameter optimization: a spectral approach 7.0
A Deep Reinforced Model for Abstractive Summarization 7.0
DCN+: Mixed Objective And Deep Residual Coattention for Question Answering 7.0
Learning Wasserstein Embeddings 7.0
Deep Learning with Logged Bandit Feedback 7.0
GraphVAE: Towards Generation of Small Graphs Using Variational Autoencoders 7.0
Fixing Weight Decay Regularization in Adam 7.0
On the Expressive Power of Overlapping Architectures of Deep Learning 7.0
Understanding Short-Horizon Bias in Stochastic Meta-Optimization 7.0
Relational Neural Expectation Maximization 7.0
Characterizing Adversarial Subspaces Using Local Intrinsic Dimensionality 7.0
When is a Convolutional Filter Easy to Learn? 7.0
Multi-level Residual Networks from Dynamical Systems View 7.0
Neural Map: Structured Memory for Deep Reinforcement Learning 7.0
附中文名：
有理对抗训练证明的分布鲁棒性
论亚当与外在的融合
通过训练递归神经网络来执行空间定位的类网格表示的出现
用于资源有效图像分类的多尺度密集网络
用混合张量分解推广扩张卷积网络
Wasserstein自动编码器
i-RevNet：深度可逆网络
学习用图表来表示程序
在非平稳和竞争的环境中通过元学习的连续适应
用预测方法稳定对抗网络
DIRT-T的无监督域适应方法
递归神经网络的交替多位量化
球形CNN | OpenReview
用于学习大群体最优行为策略的深度均值场博弈
FusionNet：通过注意力的融合和应用到机器理解
分布式优先体验重播
用于神经网络的光谱归一化边界的PAC-贝叶斯方法
无监督机器翻译仅使用单语语料库
因果GAN：学习与敌对训练的因果隐式生成模型
通过交叉渐进式训练将各个区域进行广义化
通过Skim-RNN进行神经速度读取
学习差异性私人循环语言模型
合成与自然噪声破解神经机器翻译
用景观设计学习单隐层神经网络
用于条件程序生成的神经素描学习
统一视觉环境下的模块化持续学习
AmbientGAN：来自有损测量的生成模型
深度自动编码高斯混合模型无监督异常检测
基于决策的敌对攻击：对黑盒机器学习模型的可靠攻击
跳过连接消除奇点
通过深层次的继承表征来发现特征选择
学习如何解释神经网络：模式网络和模式归因
生成对抗网络的谱归一化
掌握地下城：机械Turker下降基础的语言学习
PixelNN：基于实例的图像合成
通过深度合成代码学习压缩Word嵌入
变分网络量化
学习深度声音识别的课堂练习
深入学习物理过程：结合以前的科学知识
关于一般化的单一方向的重要性
培训RNN像CNN一样快
变分图像压缩与规模hyperprior
作为混合凸组合优化问题的深度学习
空间变换敌对的例子
攻击二值化神经网络
视觉地面想象的生成模型
政策梯度方法与行动相关基线的方差减少
高效的稀疏Winograd卷积神经网络
用乐观的方式培训GAN
无监督深度域自适应的最小熵相关校准
N2N学习：通过策略梯度强化学习的网络到网络压缩
用神经网络模拟动作动力学
素描图的神经表示
对抗案例的认证防御
终身学习与动态扩展网络
Sobolev GAN | OpenReview
反向传播通过虚空：优化控制变量的黑盒梯度估计
联合学习句法和词汇的神经语言建模
用Gumbel-Sinkhorn网络学习潜在排列
零射击视觉模仿
内在动机与自动课程的不对称性
规范和优化LSTM语言模型
主动神经定位
学习教学
用STEIN控制变量进行样本有效的政策优化
哈密尔顿型蒙特卡罗神经网络的推广
时间差分在深度强化学习中的作用分析
神经网络的可伸缩拉普拉斯逼近
超参数优化：一种光谱方法
抽象概括的深层次增强模型
DCN +：混合的目标和深刻的解决问题的答案
学习Wasserstein嵌入
深度学习与记录的强盗反馈
GraphVAE：使用变分自动编码器来生成小图
在Adam中确定权重衰减正则化
论深度学习重叠式建筑的表现力
理解随机元优化中的短视场偏差
关系神经期望最大化
利用局部固有维数表征对抗子空间
卷积滤波器什么时候易于学习？
来自动态系统视图的多级剩余网络
神经映射：深度强化学习的结构化记忆
2万+
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
