NeurIPS 2019 Events with Videos

Orals

• Updates of Equilibrium Prop Match Gradients of Backprop Through Time in an RNN with Static Input
• Uniform convergence may be unable to explain generalization in deep learning
• Causal Confusion in Imitation Learning
• Parameter elimination in particle Gibbs sampling
• Scene Representation Networks: Continuous 3D-Structure-Aware Neural Scene Representations
• Brain-Like Object Recognition with High-Performing Shallow Recurrent ANNs

Posters

• Empirically Measuring Concentration: Fundamental Limits on Intrinsic Robustness
• On Fenchel Mini-Max Learning
• Fooling Neural Network Interpretations via Adversarial Model Manipulation
• Sampling Networks and Aggregate Simulation for Online POMDP Planning
• A Simple Baseline for Bayesian Uncertainty in Deep Learning
• Communication-efficient Distributed SGD with Sketching
• Exact Rate-Distortion in Autoencoders via Echo Noise
• Explanations can be manipulated and geometry is to blame
• Chasing Ghosts: Instruction Following as Bayesian State Tracking
• Reverse engineering recurrent networks for sentiment classification reveals line attractor dynamics
• Uniform convergence may be unable to explain generalization in deep learning
• Solving Interpretable Kernel Dimensionality Reduction
• Model Compression with Adversarial Robustness: A Unified Optimization Framework
• Updates of Equilibrium Prop Match Gradients of Backprop Through Time in an RNN with Static Input
• Multi-marginal Wasserstein GAN
• Identification of Conditional Causal Effects under Markov Equivalence
• Differentiable Cloth Simulation for Inverse Problems
• A Game Theoretic Approach to Class-wise Selective Rationalization
• Correlation clustering with local objectives
• Visualizing the PHATE of Neural Networks
• Adversarial Examples Are Not Bugs, They Are Features
• A Prior of a Googol Gaussians: a Tensor Ring Induced Prior for Generative Models
• Adversarial training for free!
• Neural networks grown and self-organized by noise
• Successor Uncertainties: Exploration and Uncertainty in Temporal Difference Learning
• Point-Voxel CNN for Efficient 3D Deep Learning
• Legendre Memory Units: Continuous-Time Representation in Recurrent Neural Networks
• Learning Positive Functions with Pseudo Mirror Descent
• When to Trust Your Model: Model-Based Policy Optimization
• Hindsight Credit Assignment
• Residual Flows for Invertible Generative Modeling
• Painless Stochastic Gradient: Interpolation, Line-Search, and Convergence Rates
• Maximum Mean Discrepancy Gradient Flow
• Symmetry-Based Disentangled Representation Learning requires Interaction with Environments
• The Case for Evaluating Causal Models Using Interventional Measures and Empirical Data
• Qsparse-local-SGD: Distributed SGD with Quantization, Sparsification and Local Computations
• Parameter elimination in particle Gibbs sampling
• Imitation Learning from Observations by Minimizing Inverse Dynamics Disagreement
• Integrating Markov processes with structural causal modeling enables counterfactual inference in complex systems
• Causal Confusion in Imitation Learning
• Guided Meta-Policy Search
• Limitations of the empirical Fisher approximation for natural gradient descent
• Asymmetric Valleys: Beyond Sharp and Flat Local Minima
• On Adversarial Mixup Resynthesis
• On the Utility of Learning about Humans for Human-AI Coordination
• Implicit Generation and Modeling with Energy Based Models
• Multiclass Learning from Contradictions
• Tight Dimension Independent Lower Bound on the Expected Convergence Rate for Diminishing Step Sizes in SGD
• TAB-VCR: Tags and Attributes based Visual Commonsense Reasoning Baselines
• Joint-task Self-supervised Learning for Temporal Correspondence
• Learning Robust Options by Conditional Value at Risk Optimization
• Prediction of Spatial Point Processes: Regularized Method with Out-of-Sample Guarantees
• From voxels to pixels and back: Self-supervision in natural-image reconstruction from fMRI
• Language as an Abstraction for Hierarchical Deep Reinforcement Learning
• Structured Graph Learning Via Laplacian Spectral Constraints
• Calibration tests in multi-class classification: A unifying framework
• Likelihood Ratios for Out-of-Distribution Detection
• Multiview Aggregation for Learning Category-Specific Shape Reconstruction
• Learning Macroscopic Brain Connectomes via Group-Sparse Factorization
• Deep Multi-State Dynamic Recurrent Neural Networks Operating on Wavelet Based Neural Features for Robust Brain Machine Interfaces
• Disentangled behavioural representations
• Third-Person Visual Imitation Learning via Decoupled Hierarchical Controller
• Scene Representation Networks: Continuous 3D-Structure-Aware Neural Scene Representations
• Beyond temperature scaling: Obtaining well-calibrated multi-class probabilities with Dirichlet calibration
• Fast structure learning with modular regularization
• Learning-In-The-Loop Optimization: End-To-End Control And Co-Design Of Soft Robots Through Learned Deep Latent Representations
• Multiclass Performance Metric Elicitation
• Learning Conditional Deformable Templates with Convolutional Networks
• Modelling heterogeneous distributions with an Uncountable Mixture of Asymmetric Laplacians
• Multiway clustering via tensor block models
• Uncertainty-based Continual Learning with Adaptive Regularization
• Function-Space Distributions over Kernels
• Image Synthesis with a Single (Robust) Classifier
• Optimal Sparse Decision Trees
• Fast, Provably convergent IRLS Algorithm for p-norm Linear Regression
• Few-shot Video-to-Video Synthesis
• MCP: Learning Composable Hierarchical Control with Multiplicative Compositional Policies
• Domes to Drones: Self-Supervised Active Triangulation for 3D Human Pose Reconstruction
• Convergence-Rate-Matching Discretization of Accelerated Optimization Flows Through Opportunistic State-Triggered Control
• McDiarmid-Type Inequalities for Graph-Dependent Variables and Stability Bounds
• A Unifying Framework for Spectrum-Preserving Graph Sparsification and Coarsening
• Decentralized sketching of low rank matrices
• Icebreaker: Element-wise Efficient Information Acquisition with a Bayesian Deep Latent Gaussian Model
• DeepWave: A Recurrent Neural-Network for Real-Time Acoustic Imaging
• Visual Concept-Metaconcept Learning
• Stochastic Variance Reduced Primal Dual Algorithms for Empirical Composition Optimization
• ODE2VAE: Deep generative second order ODEs with Bayesian neural networks
• Unsupervised Meta-Learning for Few-Shot Image Classification
• Poincaré Recurrence, Cycles and Spurious Equilibria in Gradient-Descent-Ascent for Non-Convex Non-Concave Zero-Sum Games
• Fast and Accurate Stochastic Gradient Estimation
• Neuropathic Pain Diagnosis Simulator for Causal Discovery Algorithm Evaluation
• Modeling Expectation Violation in Intuitive Physics with Coarse Probabilistic Object Representations
• Unsupervised Emergence of Egocentric Spatial Structure from Sensorimotor Prediction
• The Impact of Regularization on High-dimensional Logistic Regression
• Generalization in Reinforcement Learning with Selective Noise Injection and Information Bottleneck
• Integrating Bayesian and Discriminative Sparse Kernel Machines for Multi-class Active Learning
• This Looks Like That: Deep Learning for Interpretable Image Recognition
• Imitation-Projected Programmatic Reinforcement Learning
• Reducing Noise in GAN Training with Variance Reduced Extragradient
• The Point Where Reality Meets Fantasy: Mixed Adversarial Generators for Image Splice Detection
• Multi-Criteria Dimensionality Reduction with Applications to Fairness
• Multi-Resolution Weak Supervision for Sequential Data
• Thompson Sampling for Multinomial Logit Contextual Bandits
• Nonlinear scaling of resource allocation in sensory bottlenecks
• Control What You Can: Intrinsically Motivated Task-Planning Agent
• Online Continual Learning with Maximal Interfered Retrieval
• Adversarial Training and Robustness for Multiple Perturbations
• Gossip-based Actor-Learner Architectures for Deep Reinforcement Learning
• Brain-Like Object Recognition with High-Performing Shallow Recurrent ANNs
• The spiked matrix model with generative priors
• Efficiently avoiding saddle points with zero order methods: No gradients required
• Neural Similarity Learning
• Practical Two-Step Lookahead Bayesian Optimization
• PowerSGD: Practical Low-Rank Gradient Compression for Distributed Optimization
• Seeing the Wind: Visual Wind Speed Prediction with a Coupled Convolutional and Recurrent Neural Network
• Computational Mirrors: Blind Inverse Light Transport by Deep Matrix Factorization
• Recurrent Space-time Graph Neural Networks
• Write, Execute, Assess: Program Synthesis with a REPL
• Universality in Learning from Linear Measurements
• Learning to Predict 3D Objects with an Interpolation-based Differentiable Renderer
• The Synthesis of XNOR Recurrent Neural Networks with Stochastic Logic
• Understanding the Representation Power of Graph Neural Networks in Learning Graph Topology
• On the Downstream Performance of Compressed Word Embeddings
• Piecewise Strong Convexity of Neural Networks
• Ask not what AI can do, but what AI should do: Towards a framework of task delegability
• ObjectNet: A large-scale bias-controlled dataset for pushing the limits of object recognition models
• Search-Guided, Lightly-Supervised Training of Structured Prediction Energy Networks
• GENO -- GENeric Optimization for Classical Machine Learning
• A Zero-Positive Learning Approach for Diagnosing Software Performance Regressions
• STAR-Caps: Capsule Networks with Straight-Through Attentive Routing
• Online Forecasting of Total-Variation-bounded Sequences
• Learning Deep Bilinear Transformation for Fine-grained Image Representation

Spotlights

• Legendre Memory Units: Continuous-Time Representation in Recurrent Neural Networks
• Identification of Conditional Causal Effects under Markov Equivalence
• Point-Voxel CNN for Efficient 3D Deep Learning
• Adversarial Examples Are Not Bugs, They Are Features
• Empirically Measuring Concentration: Fundamental Limits on Intrinsic Robustness
• Imitation Learning from Observations by Minimizing Inverse Dynamics Disagreement
• Asymmetric Valleys: Beyond Sharp and Flat Local Minima
• Implicit Generation and Modeling with Energy Based Models
• Residual Flows for Invertible Generative Modeling
• Guided Meta-Policy Search
• Learning Positive Functions with Pseudo Mirror Descent
• Hindsight Credit Assignment
• Calibration tests in multi-class classification: A unifying framework
• Fast structure learning with modular regularization
• Poincaré Recurrence, Cycles and Spurious Equilibria in Gradient-Descent-Ascent for Non-Convex Non-Concave Zero-Sum Games
• McDiarmid-Type Inequalities for Graph-Dependent Variables and Stability Bounds
• Optimal Sparse Decision Trees
• This Looks Like That: Deep Learning for Interpretable Image Recognition
• Adversarial Training and Robustness for Multiple Perturbations
• Multi-Criteria Dimensionality Reduction with Applications to Fairness
• On the Downstream Performance of Compressed Word Embeddings
• Ask not what AI can do, but what AI should do: Towards a framework of task delegability

Report issues here.