Efficiently bootstraps massive distributed data without over-resampling.
problem Statistical inference for massive distributed data.
method Distributed Bootstrap applied to gradients from worker machines.
result Proves optimal statistical efficiency with minimal communication.
New algorithms for fair data summarization in massive data models.
problem Fair k k k -center problem in massive data. method Streaming and distributed algorithms with provable guarantees.
result First distributed algorithm with constant approximation ratio.
DeepCMC compresses CSI for massive MIMO systems, reducing overhead and improving performance.
problem High CSI overhead in massive MIMO systems limits spectral efficiency.
method Deep learning-based fully convolutional neural network with residual layers and entropy coding.
result DeepCMC outperforms state-of-the-art schemes in CSI reconstruction quality for the same compression rate.
Improved predictions for rare labels using neural networks and ontologies.
problem Long-tailed frequency distribution in multi-label prediction problems.
method Modified neural network output layer with a Bayesian network of sigmoids leveraging ontology relationships.
result Significant improvements in per-label AUROC and average precision for less common labels.
This paper introduces a new framework for collective online learning of Gaussian processes in massive multi-agent systems.
problem The inefficiency of centralized communication in distributed machine learning systems.
method A novel Collective Online Learning Gaussian Process framework that allows each agent to build its local model and exchange it with others via peer-to-peer communication.
result Empirical results demonstrate the efficiency of the framework on both synthetic and real-world datasets.
A new asynchronous distributed EM algorithm for massive data.
problem Time-consuming E step in EM algorithms for massive data.
method Asynchronous and distributed EM (DEM) algorithm.
result Significantly faster than competing EM-type algorithms.
ADDA framework speeds up data augmentation in massive data settings.
problem Slow data augmentation in massive data settings.
method Develops asynchronous and distributed data augmentation (ADDA) framework.
result ADDA significantly speeds up data augmentation compared to parent DA algorithms.
We prove a Goldberg-Sachs theorem in dimension three. To be precise, given a three-dimensional Lorentzian manifold satisfying the topological massive gravity equations, we provide necessary and sufficient conditions on the tracefree Ricci tensor for the existence of a null line distribution whose orthogonal complement …
Paper proposes a GPU-based system for training massive deep learning models in ads systems.
problem Training massive deep learning models with terabyte-scale parameters in ads systems.
method Hierarchical GPU parameter server with 3-layer storage (GPU High-Bandwidth Memory, CPU main memory, SSD).
result 4-node hierarchical GPU parameter server trains a model 2X faster than a 150-node in-memory system.
Develops a new theory for approximating functions on massive data.
problem Challenges in machine learning with massive data.
method eignets theory for local, stratified approximation.
result Solves inverse problems like finding data probability law and function smoothness.
We present a Bayesian non-negative tensor factorization model for count-valued tensor data, and develop scalable inference algorithms (both batch and online) for dealing with massive tensors. Our generative model can handle overdispersed counts as well as infer the rank of the decomposition. Moreover, leveraging a repa…
Neural MMO simulates MMOs to study multiagent intelligence.
problem Limited research environments for multiagent intelligence.
method Developed a new game environment inspired by MMOs.
result Standard methods can learn interesting behaviors in MMOs.
A new protocol for private averaging protects data privacy in a crowd of users.
problem Protecting privacy in a crowd of users sharing personal data.
method Massively distributed algorithm for private averaging with malicious adversaries.
result Privacy is preserved even with malicious users, and the algorithm can find arbitrary accuracy solutions.
Asynchronous SVI scales up Bayesian computation for massive data.
problem Scaling stochastic variational inference to massive data.
method Lock-free asynchronous parallel implementation of SVI.
result Linear speed-up and asymptotic convergence rate O ( 1 / ( T ) ) O(1/\sqrt(T)) O ( 1/ ( T )) . DFRot improves LLMs by reducing outlier and massive activation effects.
problem Reducing outlier and massive activation effects in rotated LLMs.
method Weighted loss function and orthogonal Procrustes transforms for rotation matrix refinement.
result DFRot achieves dual free (Outlier-Free and Massive Activation-Free) with significant improvements in perplexity.
A new distributed method speeds up sparse model training.
problem Efficiently training models with massive samples and high-dimensional features.
method Distributed Dynamic Safe Screening (DDSS) method for sparsity regularized models.
result Achieves linear convergence rate and eliminates almost all inactive features.
This paper optimizes subsampling for large datasets using Poisson distribution.
problem Efficiently subsample large datasets for quasi-likelihood estimation.
method Derives optimal Poisson subsampling probabilities and develops a distributed subsampling framework.
result Consistent and asymptotically normal estimators are obtained.
Deep learning reduces training overhead in massive MIMO systems.
problem Reducing training overhead in massive MIMO systems.
method Use of deep learning (NNs) to improve CSI acquisition and feedback processes.
result Significant improvements in performance and reduced complexity.
ASHA optimizes hyperparameters for large models in parallel.
problem Large hyperparameter spaces and long training times in modern learning models.
method ASHA algorithm exploiting parallelism and early-stopping.
result ASHA outperforms existing methods and scales linearly with workers.
Massive fermions help understand index theorems without chiral symmetry.
problem Understanding index theorems in massive fermion systems.
method Reformulate chiral anomaly and index theorems with massive Dirac operators.
result Nontrivial mathematical relations between massless and massive fermions.
Stochastic Gradient Descent (SGD) has become one of the most popular optimization methods for training machine learning models on massive datasets. However, SGD suffers from two main drawbacks: (i) The noisy gradient updates have high variance, which slows down convergence as the iterates approach the optimum, and (ii)…
New framework improves fraud prediction with incremental data balancing for massive data streams.
problem Class imbalance problem in massive imbalanced data streams.
method Incremental data balancing framework using Racing Algorithm for automated balancing and Random Forest for classification.
result Better results than Batch mode on European Credit Card dataset.
SIMD operations boost Bayesian computations up to 6x faster.
problem Expensive Bayesian computations are computationally intensive and parallelizable.
method Demonstrated the utility of SIMD operations for Bayesian applications using standard libraries.
result Up to 6x improvement in floating point arithmetic performance.
In recent years, a rich variety of shrinkage priors have been proposed that have great promise in addressing massive regression problems. In general, these new priors can be expressed as scale mixtures of normals, but have more complex forms and better properties than traditional Cauchy and double exponential priors. W…
CNNs improve positioning accuracy in massive MIMO systems.
problem Improving positioning accuracy in massive MIMO systems.
method Applying CNNs to learn sparse massive MIMO channel fingerprints.
result Moderately deep CNNs achieve fractional-wavelength positioning accuracy.
Change detection in multivariate time series has applications in many domains, including health care and network monitoring. A common approach to detect changes is to compare the divergence between the distributions of a reference window and a test window. When the number of dimensions is very large, however, the naive…
Distributed computing offers a high degree of flexibility to accommodate modern learning constraints and the ever increasing size of datasets involved in massive data issues. Drawing inspiration from the theory of distributed computation models developed in the context of gradient-type optimization algorithms, we prese…
Deep nets with massive data learn spatially sparse functions.
problem Understanding the importance of massive data in deep learning.
method Established a sampling theorem and proved optimal learning rates.
result Massive data is crucial for deep nets to learn spatially sparse functions.
Random orthogonalization improves FL in massive MIMO systems without CSI.
problem Efficient model aggregation in FL with minimal channel estimation overhead.
method Combining FL with massive MIMO's channel hardening and favorable propagation, random orthogonalization reduces channel estimation overhead.
result Achieves model aggregation without CSI, significantly reducing channel estimation overhead.
Deep learning optimizes user association in Massive MIMO networks.
problem Optimizing user cell association for maximum sum-rate in Massive MIMO networks.
method Training a deep neural network to learn optimal association rules based on user positions.
result The neural network achieves the same performance as traditional optimization methods with reduced computational complexity.
COMET is a single-pass MapReduce algorithm for learning on large-scale data. It builds multiple random forest ensembles on distributed blocks of data and merges them into a mega-ensemble. This approach is appropriate when learning from massive-scale data that is too large to fit on a single machine. To get the best acc…
Develops an online nonparametric classifier for massive data.
problem Challenges of batch kernel-based nonparametric classifiers in massive data.
method Online principle components analysis to reduce dimensionality, followed by stochastic approximation algorithm for real-time calculation.
result Online classifier provides the best trade-off between accuracy and computation cost.
Infinite Tucker Decomposition (InfTucker) and random function prior models, as nonparametric Bayesian models on infinite exchangeable arrays, are more powerful models than widely-used multilinear factorization methods including Tucker and PARAFAC decomposition, (partly) due to their capability of modeling nonlinear rel…
Study evaluates topological contributions in massive SQCD on compact 4-manifolds.
problem Analyzing topological path integrals for massive SQCD with up to 3 massive hypermultiplets.
method Decouples hypermultiplets, evaluates massless limit, and merges singularities at Argyres-Douglas points. Uses mass expansions for P 2 \mathbb{P}^2 P 2 and K 3 K3 K 3 . result Physical partition functions match mathematical results on Segre numbers of instanton moduli spaces.
Unified framework for photon and massive particle hypersurfaces in stationary spacetimes.
problem Understanding photon and massive particle hypersurfaces in stationary spacetimes.
method Unified framework using Killing-invariant timelike hypersurfaces and associated Finsler structures.
result Conditions for a hypersurface to be a photon or massive particle hypersurface are established.
In this work we simulate null geodesics for the Bonnor massive dipole metric by implementing a symbolic-numerical algorithm in Sage and Python. This program is also capable of visualizing in 3D, in principle, the geodesics for any given metric. Geodesics are launched from a common point, collectively forming a cone of …
The paper studies a method to sample nodes from a massive graph using personalized PageRank.
problem Sampling from a massive network is expensive and impractical; the paper provides an alternative.
method The paper introduces a crawling method to approximate the personalized PageRank vector without querying the entire graph.
result The adjusted personalized PageRank vector can effectively select nodes within the same block as the seed node.
Introduces a massive variant of Ray-Singer Torsion to avoid zero modes in topological field theories.
problem Avoiding zero modes in the evaluation of path integrals for topological field theories.
method Introduces a massive variant of the Ray-Singer Torsion, involving determinants of the twisted Laplacian with mass but without zero modes.
result Explicitly evaluates the massive Ray-Singer Torsion on product manifolds and mapping tori.
New method for efficient inference in large datasets.
problem Statistical inference in massive datasets.
method Combines divide-and-conquer method and empirical likelihood.
result Reduces computation burden and demonstrates effectiveness.
QEM uses parallel importance weighting for fast approximate Bayesian inference.
problem Bayesian inference challenges in large models with many observations and latent variables.
method Expectation Maximization (EM) with massively parallel importance weighting.
result QEM is faster and more scalable than RWS and VI.
We consider the problem of learning classifiers for labeled data that has been distributed across several nodes. Our goal is to find a single classifier, with small approximation error, across all datasets while minimizing the communication between nodes. This setting models real-world communication bottlenecks in the …
Deep learning reduces noise in weak lensing mass maps using GANs.
problem Noise reduction in weak lensing mass maps.
method Generative adversarial networks (GANs) applied to Subaru Hyper Suprime-Cam data.
result GANs successfully reproduce non-Gaussian information in denoised maps, showing stronger cosmological dependence.
The paper addresses frequency-dependent distortions in massive MIMO systems and proposes a method to recover covariance matrices.
problem Frequency-dependent distortions in the covariance matrix of massive MIMO systems.
method Proposes a novel UL-DL covariance interpolation technique under a mild reciprocity condition.
result The proposed method can recover the covariance matrix in the DL from an estimate in the UL, especially in FDD massive MIMO systems.
Developed PathInf for network inference from incomplete data.
problem Massive and non-uniformly distributed missing values in data.
method Two-stage inference model: data summarization and graph inference.
result Consistently superior performance compared to state-of-the-art methods.
Study improves scalability of cell-free massive MIMO networks by optimizing UE-AP association.
problem Optimizing UE-AP association in cell-free massive MIMO networks.
method Deep learning algorithm using Bidirectional Long Short-Term Memory cells and hybrid probabilistic weight updating.
result Enhanced scalability without retraining, robust against pilot contamination.
Paper corrects deep learning for noisy labels.
problem Overfitting to imperfectly labeled data.
method Distribution correction approach to handle noisy inputs.
result Significantly higher accuracy compared to alternative methods.
New method samples manifolds efficiently using Dirichlet distribution.
problem Sampling on complex manifolds efficiently.
method Data-driven Dirichlet sampling on manifolds.
result Efficient sampling respects manifold structure with low computational effort.
This paper decentralizes SVI for massive data using ADMM.
problem Efficient approximate inference for massive data with complex models.
method ADMM-based distributed SVI algorithm.
result Decentralized SVI improves performance and robustness.