arXiv 2012

13,904 papers
In 2012, arXiv hosted 587 papers across diverse computational and mathematical disciplines. A central focus was placed on numerical methods and optimization, particularly structure-preserving geometric numerical integration and implicit-explicit time integration for multi-scale physical systems. Research also addressed computational challenges in high-dimensional data, including scaling Gaussian processes beyond cubic complexity and learning from low-sample-size biological datasets. Notable work explored the intersection of Bayesian inference and optimization, as well as hierarchical variance reduction through multilevel Monte Carlo methods. The year's contributions extended into statistical regression and signal processing, with studies on robust variable selection and sub-Nyquist imaging via sparse reconstruction. Research in stochastic systems was prominent, covering online optimization of stochastic hybrid systems and performance evaluations of pedestrian tracking algorithms. Prolific contributors included Michael Jordan, Andrew Stuart, and Terry L. Friesz. Other key areas of investigation involved modeling the missing heritability of complex traits and analyzing spectrum sharing systems over fading channels. Collectively, these works reflected a trend toward improving the efficiency and robustness of algorithms in complex, high-dimensional environments.

Top Research Directions

Research directions most represented by arXiv 2012 papers.

Learning Densities with Intractable Normalizing Constants 46
Convergence of Message Passing on Loopy Graphs 43
Variational Inference as an Optimization Problem 42
The Empirical Validation Gap 40
How Network Topology Governs Decentralized Learning Convergence 33
Stationary Point Convergence in Nonconvex Optimization 33
Scaling Gaussian Processes Beyond Cubic Complexity 32
Semidefinite Relaxation for Non-Convex Engineering Optimization 28
Incremental Maintenance of Dynamic Graph Indices 27
Overlapping Community Detection in Complex Networks 24
Aligning Selfish Routing with Socially Optimal Traffic Flow 24
Kernel Hyperparameter Sensitivity in Distributional Metrics 24
Overcoming the Cubic Complexity of Kernel Methods 24
Generalizing Provable Control to Nonlinear Systems 22
Self Supervised Regularization for Sparse Recommender Systems 22

Top Authors

AuthorPapers
Yoshua Bengio20
Mohamed-Slim Alouini18
Michael Jordan17
Daniel Hsu14
Adnan Darwiche13
Bernhard Schölkopf13
Max Welling13
Craig Boutilier12
Eric P. Xing12
Feng Xia12
Michael Rabbat12
Rong Jin12
C. Seshadhri11
Faramarz Fekri11
Ness Shroff11
Xiang-Yang Li11
Yuantao Gu11
Georgios B. Giannakis10
Jürgen Schmidhuber10
Peter Spirtes10

Impactful Papers

Papers from arXiv 2012 that shaped research clusters.

Foundational

Influential

Papers

Title
Convergence analysis of the FOCUSS algorithm
Convergence of multivariate belief propagation, with applications to cuckoo hashing and load balancing
Convergence of the Continuous Time Trajectories of Isotropic Evolution Strategies on Monotonic C^2-composite Functions
Convergence of the EM Algorithm for Gaussian Mixtures with Unbalanced Mixing Coefficients
Convergence Rates for Differentially Private Statistical Estimation
Convergence Rates of Biased Stochastic Optimization for Learning Sparse Ising Models
Convergence Speed of Binary Interval Consensus
Convergent filtered schemes for the Monge-Ampère partial differential equation
Convergent message passing algorithms - a unifying view
Convergent Message-Passing Algorithms for Inference over General Graphs with Convex Free Energies
Convergent Propagation Algorithms via Oriented Trees
Converting online algorithms to local computation algorithms
Convex Algorithms for Nonnegative Matrix Factorization
Convex Coding
Convexifying the Bethe Free Energy
Convexity Conditions for 802.11 WLANs
Convex Multitask Learning with Flexible Task Clusters
Convex Point Estimation using Undirected Bayesian Transfer Hierarchies

« prev Page 75/279 next »