ICML 2018 accepted papers on Learning theory and algorithms
149 of the 623 papers accepted at ICML 2018 (23.9%) are on Learning theory and algorithms, compared with 26.9% at ICML 2017. They fall into 9 topics.
See ICML 2018 in the interactive map
Submitting to ICML? Checklist →
Working on something in this area? Describe your idea in scime atlas to see the closest papers by meaning and how crowded the spot has become.
Topics and papers
Prediction and decision losses · 31 papers
- Representation Tradeoffs for Hyperbolic Embeddings (190 citations)
- Deep Linear Networks with Arbitrary Loss: All Local Minima Are Global (80 citations)
- An Optimal Control Approach to Deep Learning and Applications to Discrete-Weight Neural Networks (48 citations)
- Open Category Detection with PAC Guarantees (46 citations)
- Understanding the Loss Surface of Neural Networks for Binary Classification (45 citations)
- SAFFRON: an Adaptive Algorithm for Online Control of the False Discovery Rate (34 citations)
- Multi-Fidelity Black-Box Optimization with Hierarchical Partitions (28 citations)
- QuantTree: Histograms for Change Detection in Multivariate Data Streams (24 citations)
- Binary Partitions with Approximate Minimum Impurity (22 citations)
- Finding Influential Training Samples for Gradient Boosted Decision Trees (19 citations)
- Stein Points (19 citations)
- The Multilinear Structure of ReLU Networks (11 citations)
and 19 more in the interactive map.
Matrix and tensor methods · 27 papers
- Deep Models of Interactions Across Sets (73 citations)
- Canonical Tensor Decomposition for Knowledge Base Completion (48 citations)
- SQL-Rank: A Listwise Approach to Collaborative Ranking (31 citations)
- A Primal-Dual Analysis of Global Optimality in Nonconvex Low-Rank Matrix Recovery (22 citations)
- prDeep: Robust Phase Retrieval with a Flexible Deep Network (19 citations)
- Signal and Noise Statistics Oblivious Orthogonal Matching Pursuit (12 citations)
- Gradient Descent for Sparse Rank-One Matrix Completion for Crowd-Sourced Aggregation of Sparsely Interacting Workers (11 citations)
- Subspace Embedding and Linear Regression with Orlicz Norm (11 citations)
- Near Optimal Frequent Directions for Sketching Dense and Sparse Matrices (10 citations)
- Parameterized Algorithms for the Matrix Completion Problem (10 citations)
- WHInter: A Working set algorithm for High-dimensional sparse second order Interaction models (9 citations)
- Low-Rank Riemannian Optimization on Positive Semidefinite Stochastic Matrices with Applications to Graph Clustering (8 citations)
and 15 more in the interactive map.
Kernels and regression theory · 19 papers
- Trainable Calibration Measures for Neural Networks from Kernel Mean Embeddings (104 citations)
- To Understand Deep Learning We Need to Understand Kernel Learning (100 citations)
- Meta-Learning by Adjusting Priors Based on Extended PAC-Bayes Theory (95 citations)
- Stronger Generalization Bounds for Deep Nets via a Compression Approach (81 citations)
- An Iterative, Sketching-based Framework for Ridge Regression (25 citations)
- Learning and Memorization (19 citations)
- Dropout Training, Data-dependent Regularization, and Generalization Bounds (17 citations)
- Optimal Tuning for Divide-and-conquer Kernel Ridge Regression with Massive Data (16 citations)
- Approximate Leave-One-Out for Fast Parameter Tuning in High Dimensions (13 citations)
- Nonoverlap-Promoting Variable Selection (6 citations)
- On the Spectrum of Random Features Maps of High Dimensional Data (4 citations)
- Adaptive Sampled Softmax with Kernel Based Sampling (3 citations)
and 7 more in the interactive map.
Sample complexity · 19 papers
- SBEED: Convergent Reinforcement Learning with Nonlinear Function Approximation (137 citations)
- Accelerated Spectral Ranking (20 citations)
- Spatio-temporal Bayesian On-line Changepoint Detection with Model Selection (17 citations)
- Rates of Convergence of Spectral Methods for Graphon Estimation (16 citations)
- Crowdsourcing with Arbitrary Adversaries (15 citations)
- Nearly Optimal Robust Subspace Tracking (15 citations)
- Learning Diffusion using Hyperparameters (14 citations)
- Randomized Block Cubic Newton Method (13 citations)
- Distributed Nonparametric Regression under Communication Constraints (8 citations)
- Composite Marginal Likelihood Methods for Random Utility Models (7 citations)
- Locally Private Hypothesis Testing (7 citations)
- Local Density Estimation in High Dimensions (6 citations)
and 7 more in the interactive map.
Clustering · 17 papers
- Submodular Hypergraphs: p-Laplacians, Cheeger Inequalities and Spectral Clustering (59 citations)
- Hierarchical Clustering with Structural Constraints (34 citations)
- CRAFTML, an Efficient Clustering-based Random Forest for Extreme Multi-label Learning (30 citations)
- Massively Parallel Algorithms and Hardness for Single-Linkage Clustering under $\ell_p$ Distances (21 citations)
- Quickshift++: Provably Good Initializations for Sample-Based Mean Shift (21 citations)
- Analysis of Minimax Error Rate for Crowdsourcing and Its Application to Worker Clustering Model (18 citations)
- SMAC: Simultaneous Mapping and Clustering Using Spectral Decompositions (17 citations)
- Partial Optimality and Fast Lower Bounds for Weighted Correlation Clustering (10 citations)
- Distributed Clustering via LSH Based Data Partitioning (8 citations)
- K-means clustering using random matrix sparsification (7 citations)
- Spectrally Approximating Large Graphs with Smaller Graphs (7 citations)
- Theoretical Analysis of Sparse Subspace Clustering with Missing Entries (5 citations)
and 5 more in the interactive map.
Submodular and game algorithms · 12 papers
- Beyond 1/2-Approximation for Submodular Maximization on Massive Data Streams (42 citations)
- Scalable Deletion-Robust Submodular Maximization: Data Summarization with Privacy and Fairness Constraints (37 citations)
- Data Summarization at Scale: A Two-Stage Submodular Approach (30 citations)
- Greed is Still Good: Maximizing Monotone Submodular+Supermodular (BP) Functions (27 citations)
- Proportional Allocation: Simple, Distributed, and Diverse Matching with High Entropy (17 citations)
- Fast Maximization of Non-Submodular, Monotonic Functions on the Integer Lattice (16 citations)
- Approximation Guarantees for Adaptive Sampling (15 citations)
- Decentralized Submodular Maximization: Bridging Discrete and Continuous Settings (12 citations)
- Learning to Optimize Combinatorial Functions (11 citations)
- Budgeted Experiment Design for Causal Structure Learning (9 citations)
- Weakly Submodular Maximization Beyond Cardinality Constraints: Does Randomization Help Greedy? (4 citations)
- Constrained Interacting Submodular Groupings (2 citations)
Fairness · 11 papers
- Preventing Fairness Gerrymandering: Auditing and Learning for Subgroup Fairness (383 citations)
- Fairness Without Demographics in Repeated Loss Minimization (211 citations)
- Learning Adversarially Fair and Transferable Representations (184 citations)
- Delayed Impact of Fair Machine Learning (163 citations)
- A Reductions Approach to Fair Classification (103 citations)
- Multicalibration: Calibration for the (Computationally-Identifiable) Masses (100 citations)
- Residual Unfairness in Fair Machine Learning from Prejudiced Data (79 citations)
- Nonconvex Optimization for Regression with Fairness Constraints (60 citations)
- Fair and Diverse DPP-Based Data Summarization (54 citations)
- Probably Approximately Metric-Fair Learning (37 citations)
- Blind Justice: Fairness with Encrypted Sensitive Attributes (31 citations)
Combinatorial and search optimization · 11 papers
- Towards Fast Computation of Certified Robustness for ReLU Networks (149 citations)
- Differentiable Dynamic Programming for Structured Prediction and Attention (78 citations)
- Learning to Branch (68 citations)
- Learning to search with MCTSnets (26 citations)
- Fast Bellman Updates for Robust MDPs (19 citations)
- Selecting Representative Examples for Program Synthesis (14 citations)
- LeapsAndBounds: A Method for Approximately Optimal Algorithm Configuration (8 citations)
- Orthogonality-Promoting Distance Metric Learning: Convex Relaxation and Theoretical Analysis (5 citations)
- Celer: a Fast Solver for the Lasso with Dual Extrapolation (2 citations)
- Compiling Combinatorial Prediction Games (2 citations)
- Safe Element Screening for Submodular Function Minimization
Optimal transport · 2 papers
- Stochastic Wasserstein Barycenters (46 citations)
- A Two-Step Computation of the Exact GAN Wasserstein Distance (23 citations)
