Hot Topics at ICML 2018
Too many abstracts to browse?
Here is a list of topics and related talks, automatically extracted with data science
learning transfer
- Learning with Abandonment
- Learning to Branch
- Transfer Learning via Learning to Transfer
- Online Learning with Abstentions
- Active Learning with Logged Data
- ContextNet: Deep learning for Star Galaxy Classification
- To Understand Deep Learning We Need to Understand Kernel Learning
- Curriculum Learning by Transfer Learning: Theory and Experiments with Deep Networks
- Been There, Done That: Meta-Learning with Episodic Recall
- PDE-Net: Learning PDEs from Data
neural network
- Born Again Neural Networks
- Efficient Neural Audio Synthesis
- Orthogonal Recurrent Neural Networks with Scaled Cayley Transform
- Conditional Neural Processes
- Differentiable plasticity: training plastic neural networks with backpropagation
- Learning Compact Neural Networks with Regularization
- Distilling the Posterior in Bayesian Neural Networks
- Extracting Automata from Recurrent Neural Networks Using Queries and Counterexamples
- DCFNet: Deep Neural Network with Decomposed Convolutional Filters
- Stabilizing Gradients for Deep Neural Networks via Efficient SVD Parameterization
gradient stochastic
- Escaping Saddles with Stochastic Gradients
- SADAGRAD: Strongly Adaptive Stochastic Gradient Methods
- Global Convergence of Policy Gradient Methods for Linearized Control Problems
- Asynchronous Decentralized Parallel Stochastic Gradient Descent
- Data-Dependent Stability of Stochastic Gradient Descent
- Projection-Free Online Optimization with Stochastic Gradient: From Convexity to Submodularity
- An Inference-Based Policy Gradient Method for Learning Options
- Communication Efficient Gradient Coding
- On the Theory of Variance Reduction for Stochastic Gradient Monte Carlo
- Stein Variational Gradient Descent Without Gradient
adversarial generative
- First Order Generative Adversarial Networks
- Chi-square Generative Adversarial Network
- Tempered Adversarial Networks
- Reinforcing Adversarial Robustness using Model Confidence Induced by Adversarial Training
- Geometry Score: A Method For Comparing Generative Adversarial Networks
- Composite Functional Gradient Learning of Generative Adversarial Models
- Improved Training of Generative Adversarial Networks Using Representative Features
- Provable Defenses against Adversarial Examples via the Convex Outer Adversarial Polytope
- GraphRNN: A Deep Generative Model for Graphs
- RadialGAN: Leveraging multiple datasets to improve target-specific predictive models using Generative Adversarial Networks
reinforcement agent
- Mix & Match - Agent Curricula for Reinforcement Learning
- Continual Reinforcement Learning with Complex Synapses
- AnonLib: Enabling Composition in Distributed Reinforcement Learning
- Coordinated Exploration in Concurrent Reinforcement Learning
- Modeling Others using Oneself in Multi-Agent Reinforcement Learning
- Deep Variational Reinforcement Learning for POMDPs
- Fully Decentralized Multi-Agent Reinforcement Learning with Networked Agents
- Mean Field Multi-Agent Reinforcement Learning
- Automatic Goal Generation for Reinforcement Learning Agents
- State Abstractions for Lifelong Reinforcement Learning
variational inference
- Inference Suboptimality in Variational Autoencoders
- Semi-Implicit Variational Inference
- Vadam: Fast and Scalable Variational Inference by Perturbing Adam
- Yes, but Did It Work?: Evaluating Variational Inference
- Variational Network Inference: Strong and Stable with Concrete Support
- Deep Variational Reinforcement Learning for POMDPs
- Tighter Variational Bounds are Not Necessarily Better
- Semi-Amortized Variational Autoencoders
- Black-box Variational Inference for Stochastic Differential Equations
- Noisy Natural Gradient as Variational Inference
deep model
- ContextNet: Deep learning for Star Galaxy Classification
- GraphRNN: A Deep Generative Model for Graphs
- Tropical Geometry of Deep Learning
- PredRNN++: Towards A Resolution of the Deep-in-Time Dilemma in Spatiotemporal Predictive Learning
- Calibrated Estimates of Predictive Uncertainty in Deep Learning
- To Understand Deep Learning We Need to Understand Kernel Learning
- Optimization Landscape and Expressivity of Deep CNNs
- StrassenNets: Deep learning with a multiplication budget
- Constraining the Dynamics of Deep Probabilistic Models
- Can Deep Reinforcement Learning Solve Erdos-Selfridge-Spencer Games?
optimization bayesian
- BOCK : Bayesian Optimization with Cylindrical Kernels
- Policy Optimization with Demonstrations
- Distributional Optimization from Samples
- Stagewise Safe Bayesian Optimization with Gaussian Processes
- Tight Regret Bounds for Bayesian Optimization in One Dimension
- Bayesian Optimization of Combinatorial Structures
- AutoPrognosis: Automated Clinical Prognostic Modeling via Bayesian Optimization with Structured Kernel Learning
- Batched Bayesian Optimization via Multi-objective Acquisition Ensemble for Automated Analog Circuit Design
- Nonconvex Optimization for Fair Regression
- Optimization, fast and slow: optimally switching between local and Bayesian optimization
algorithm distributed
- Approximation Algorithms for Cascading Prediction Models
- Efficient Approximation Algorithms for Distributed Clustering
- A Distributed Second-Order Algorithm You Can Trust
- Parallel and Streaming Algorithms for K-Core Decomposition
- A Delay-tolerant Proximal-Gradient Algorithm for Distributed Learning
- Efficient First-Order Algorithms for Adaptive Signal Denoising
- Parameterized Algorithms for the Matrix Completion Problem
- Cut-Pursuit Algorithm for Regularizing Nonsmooth Functionals with Graph Total Variation
- Gradient Primal-Dual Algorithm Converges to Second-Order Stationary Solution for Nonconvex Distributed Optimization Over Networks
- Learning Steady-States of Iterative Algorithms over Graphs
based random
- Fitting New Speakers Based on a Short Untranscribed Sample
- Comparison-Based Random Forests
- Lipschitz Continuity in Model-based Reinforcement Learning
- Adaptive Sampled Softmax with Kernel Based Sampling
- Attention-based Deep Multiple Instance Learning
- Training Neural Machines with Trace-Based Supervision
- Fair and Diverse DPP-based Data Summarization
- Ranking Distributions based on Noisy Sorting
- Feedback-Based Tree Search for Reinforcement Learning
- Approximate message passing for amplitude based optimization
robust examples
- More Robust Doubly Robust Off-policy Evaluation
- Does Distributionally Robust Supervised Learning Give Robust Classifiers?
- Synthesizing Robust Adversarial Examples
- Nearly Optimal Robust Subspace Tracking and Dynamic Robust PCA
- DRACO: Robust Distributed Training via Redundant Gradients
- Learning to Reweight Examples for Robust Deep Learning
- Fast Bellman Updates for Robust MDPs
- Byzantine-Robust Distributed Learning: Towards Optimal Statistical Rates
- BO-HB: Robust and Efficient Hyperparameter Optimization at Scale
- Towards Binary-Valued Gates for Robust LSTM Training
using sequence
- Learning Diffusion using Hyperparameters
- Human Activity Prediction Using Sequence Earley Algorithm
- Extracting Automata from Recurrent Neural Networks Using Queries and Counterexamples
- Scalable Bilinear Pi Learning Using State and Action Features
- Modeling Others using Oneself in Multi-Agent Reinforcement Learning
- K-means clustering using random matrix sparsification
- Solving Partial Assignment Problems using Random Clique Complexes
- SMAC: Simultaneous Mapping and Clustering Using Spectral Decompositions
- Compressing Neural Networks using the Variational Information Bottelneck
- Transfer in Deep Reinforcement Learning Using Successor Features and Generalised Policy Improvement
Brought to you by Tong Wei