Study calculates tail risk for various mixture distributions.
problem Estimating tail risk for complex distribution mixtures.
method Analyzes tail conditional expectation for location-scale mixtures of elliptical distributions.
result Developed methods for calculating tail risk in various distributions.
Paper proposes MWDE for estimating finite location-scale mixtures.
problem Estimating finite location-scale mixtures using MLE is problematic.
method Investigates minimum Wasserstein distance estimators (MWDE).
result MWDE is consistent and provides a numerical solution.
Study uniform rates for estimating Gaussian mixtures without separation assumption.
problem Estimating parameters in two-component Gaussian mixtures without separation.
method Uniform convergence rates derived using minimax lower bounds and careful analysis of polynomial equalities.
result Phase transition in optimal estimation rate based on mixture balance.
Bayesian Optimization improves data mixture selection for large language models.
problem Optimizing the training data mixtures for large language models.
method Viewed as a black-box hyperparameter optimization problem, using Bayesian Optimization.
result Consistently strong results with speed-ups of over 500%.
Natural image statistics exhibit hierarchical dependencies across multiple scales. Representing such prior knowledge in non-factorial latent tree models can boost performance of image denoising, inpainting, deconvolution or reconstruction substantially, beyond standard factorial "sparse" methodology. We derive a large …
In this paper, a scale mixture of Normal distributions model is developed for classification and clustering of data having outliers and missing values. The classification method, based on a mixture model, focuses on the introduction of latent variables that gives us the possibility to handle sensitivity of model to out…
The paper studies scaling limits of Wasserstein metrics on Gaussian mixture models.
problem Understanding the scaling limits of Wasserstein metrics on Gaussian mixture models.
method Scaling limit approach on Gaussian mixture models, including inhomogeneous and extended models.
result Existence of the limit of the Wasserstein metric after renormalization for GMMs with zero variance.
Flexible model captures varying scales in data clusters.
problem Real-world data often exhibits varying scales or intensities, violating the homogeneity assumption of classical Gaussian mixture models.
method Individual-heterogeneous sub-Gaussian mixture model with an efficient spectral method for exact recovery.
result The method provably achieves exact recovery of true cluster labels under mild separation conditions.
This paper studies identifiability and convergence behaviors for parameters of multiple types in finite mixtures, and the effects of model fitting with extra mixing components. First, we present a general theory for strong identifiability, which extends from the previous work of Nguyen [2013] and Chen [1995] to address…
Paper introduces scalable clustering for large datasets with outliers.
problem Lack of scalable algorithms for large datasets with outliers.
method Provable robust clustering algorithm based on loss minimization for Gaussian mixture models.
result Algorithm provides high accuracy with theoretical guarantees and outperforms existing methods.
Improved HMoE models using Laplace gating function enhance expert specialization and performance.
problem Improving performance of hierarchical mixture of experts models.
method Used Laplace gating function instead of Softmax in hierarchical mixture of experts models.
result Laplace gating function accelerates expert convergence and enhances specialization.
Proposes scale mixture of NNGPs for more flexible stochastic processes.
problem Limited focus on broadening the class of stochastic processes from NNGPs.
method Scale mixture of NNGPs with scale priors on last-layer parameters.
result Turns neural networks into a richer class of stochastic processes.
Model detects epileptic seizures in EEG with high sensitivity.
problem Detecting epileptic seizures in EEG signals.
method Time-series scale mixture model with hidden Markov structure.
result Model outperformed baselines in seizure detection.
ELU algorithm improves on EM for over-specified Gaussian mixtures.
problem Slow convergence of EM in over-specified Gaussian mixtures.
method Developed ELU algorithm for two-component mixtures, combining exponential location update and gradient descent.
result ELU converges to final statistical radius after logarithmic iterations, resolving open question.
We propose the Bayesian bridge estimator for regularized regression and classification. Two key mixture representations for the Bayesian bridge model are developed: (1) a scale mixture of normals with respect to an alpha-stable random variable; and (2) a mixture of Bartlett--Fejer kernels (or triangle densities) with r…
The Perona-Malik model has been very successful at restoring images from noisy input. In this paper, we reinterpret the Perona-Malik model in the language of Gaussian scale mixtures and derive some extensions of the model. Specifically, we show that the expectation-maximization (EM) algorithm applied to Gaussian scale …
We propose a practical and scalable Gaussian process model for large-scale nonlinear probabilistic regression. Our mixture-of-experts model is conceptually simple and hierarchically recombines computations for an overall approximation of a full Gaussian process. Closed-form and distributed computations allow for effici…
Theory of MoE Transformers' generalization and scaling.
problem Understanding the generalization and scaling of Mixture-of-Experts (MoE) Transformers.
method Developed a theory that separates active capacity from routing combinatorics, derived a sup-norm covering-number bound, and proved a constructive approximation theorem.
result Generalization and scaling laws for MoE Transformers, showing how active capacity and routing structure affect performance.
AutoScale improves LLM pre-training by adjusting data mixtures at different scales.
problem Data mixtures that work well at small scales may not perform as well at larger scales.
method AutoScale uses a two-stage approach: fitting a model to predict loss under different compositions and extrapolating optimal compositions to larger scales.
result AutoScale accelerates convergence and improves downstream performance.
We present a probabilistic model for natural images which is based on Gaussian scale mixtures and a simple multiscale representation. In contrast to the dominant approach to modeling whole images focusing on Markov random fields, we formulate our model in terms of a directed graphical model. We show that it is able to …
We consider unsupervised estimation of mixtures of discrete graphical models, where the class variable corresponding to the mixture components is hidden and each mixture component over the observed variables can have a potentially different Markov graph structure and parameters. We propose a novel approach for estimati…
Mixture models are a fundamental tool in applied statistics and machine learning for treating data taken from multiple subpopulations. The current practice for estimating the parameters of such models relies on local search heuristics (e.g., the EM algorithm) which are prone to failure, and existing consistent methods …
Paper establishes universal lower bounds and optimal rates for clustering sub-exponential mixture models.
problem Achieving optimal error rates in clustering sub-exponential mixture models.
method Establishes universal lower bounds and demonstrates iterative algorithms' optimality in sub-exponential mixture models.
result Iterative algorithms achieve the universal lower bound in sub-exponential mixture models.
Bayesian model averaging improves causal effect estimation by averaging over multiple models.
problem Estimating causal effects under linear Structural Causal Models (SCMs).
method Bayesian model averaging using Gaussian scale mixture distributions for computational efficiency.
result Bayesian model averaging is optimal for causal effect estimation.
Paper introduces new Gromov-type distances for comparing Gaussian mixture models.
problem Comparing distributions across different metric spaces using Gromov-Wasserstein distances.
method Incorporates invariance properties into MW2, introducing MGW2 and EW2.
result MGW2 and EW2 are efficient for estimating distances between GMMs in practical applications.
Efficiently trains large GMMs with millions to billions of parameters.
problem Training large Gaussian Mixture Models (GMMs) is computationally expensive.
method Derives a variational approximation integrated with mixtures of factor analyzers (MFAs) to reduce complexity.
result Sublinear scaling in training GMMs, achieving significant speed-ups.
Generative model improves EMG pattern recognition accuracy.
problem Stochastic characteristics of EMG signals not fully considered in existing classification methods.
method Scale mixture-based stochastic generative model with variational Bayesian learning.
result Proposed method outperforms conventional classifiers in EMG pattern recognition.
A family of parsimonious shifted asymmetric Laplace mixture models is introduced. We extend the mixture of factor analyzers model to the shifted asymmetric Laplace distribution. Imposing constraints on the constitute parts of the resulting decomposed component scale matrices leads to a family of parsimonious models. An…
In this paper we propose a novel framework for the construction of sparsity-inducing priors. In particular, we define such priors as a mixture of exponential power distributions with a generalized inverse Gaussian density (EP-GIG). EP-GIG is a variant of generalized hyperbolic distributions, and the special cases inclu…
Paper analyzes approximation and learning of MoEs with (P)ReLU activation.
problem Scaling up deep learning models with MoEs and (P)ReLU activation.
method Approximation and learning-theoretic analysis of MoMLPs with (P)ReLU.
result MoMLPs can uniformly approximate Lipschitz functions with ε accuracy using O(ε−1) parameters. New algorithm learns permutations mixtures with optimal sample complexity.
problem Learning mixtures of permutations in high-dimensional settings.
method Combining groups of pairwise comparisons and combinatorial method of moments.
result Optimal sample complexity proportional to log(n) for high-dimensional data.
Diffusion models achieve high-quality samples from complex high-dimensional Gaussian mixtures without scaling with dimension.
problem Achieving accurate sampling from high-dimensional distributions using diffusion models.
method Investigates the effectiveness of diffusion models in sampling from Gaussian Mixture Models (GMMs) without scaling with dimension.
result DDPM requires at most O(1/ε) iterations to attain an ε-accurate distribution in total variation distance, independent of dimension and number of components. Proposes a new Bayesian mixture of student-t processes for modeling non-stationary data.
problem Non-stationary data with non-Gaussian errors.
method Bayesian mixture of student-t processes with an overall-local scale structure, using SMC for online inference.
result Superior performance compared to Gaussian processes on real-world data.
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…
Optimal transport between Gaussian Mixture Models improves domain adaptation efficiency.
problem Adapting machine learning models to new data distributions with minimal access.
method Optimal transport between Gaussian Mixture Models (GMMs) for domain adaptation.
result Our methods are more efficient and scalable with sample size and dimensions.
New scaling framework for MoE architectures ensures stability and optimal performance at scale.
problem Lack of principled understanding of how hyperparameters should scale in MoE architectures.
method Developed a novel Dynamical Mean Field Theory (DMFT) for three scaling regimes of MoE architectures.
result Derived Maximally Scale-Stable Parameterization (MSSP) for SGD and Adam, providing robust learning rate transfer and monotonic improvement with scale.
Improves scalability and efficiency of mixture models in black-box variational inference.
problem Scaling mixture models in black-box variational inference leads to high parameter and time costs.
method Introduces MISVAE for amortized mixture parameter space and new ELBO estimators.
result Achieves superior estimation performance with fewer parameters and shorter inference time.
SMM improves signal recovery from noisy data.
problem Estimating signals from noisy and scaled observations.
method Spiked mixture model (SMM) with EM algorithm.
result SMM outperforms GMM in signal recovery.
VMoER improves uncertainty quantification in MoE layers for scalable foundation models.
problem Uncertainty quantification in large-scale models like MoE layers.
method Structured Bayesian approach with amortized variational inference over routing logits and temperature parameter inference.
result Improves routing stability, reduces calibration error, and increases AUROC by 12%.
A new sampler speeds up Bayesian mixture models.
problem Sampling from Bayesian finite mixture models is slow and hard.
method Introduces a non-reversible sampling scheme for Bayesian finite mixture models.
result The new sampler outperforms classical samplers in many scenarios, especially during convergence.
In this paper we introduce a new approach to topic modelling that scales to large datasets by using a compact representation of the data and by leveraging the GPU architecture. In this approach, topics are learned directly from the co-occurrence data of the corpus. In particular, we introduce a novel mixture model whic…
How can we train a statistical mixture model on a massive data set? In this work we show how to construct coresets for mixtures of Gaussians. A coreset is a weighted subset of the data, which guarantees that models fitting the coreset also provide a good fit for the original data set. We show that, perhaps surprisingly…
A digital twin for multi-scale systems uses physics-based and machine learning models.
problem Lack of application-specific details in digital twin technology.
method Strategically separates into physics-based and data-driven models; uses mixture of experts with Gaussian Process.
result Robust and accurate predictions at future time-steps for multi-scale systems.
The paper optimizes model selection and parameter estimation for multi-dimensional Gaussian Mixture Models.
problem Learning and distinguishing multi-dimensional Gaussian Mixture Models with reliable model order selection and efficient estimation.
method The paper establishes an information-theoretic lower bound and proposes a thresholding-based estimation algorithm with a time complexity of O(k^2 n). It also introduces a gradient-based minimization method with PCA for high-dimensional cases.
result The proposed method matches the established lower bound in sample complexity and achieves optimal parametric convergence rate.
Higher granularity in MoE models boosts expressivity exponentially.
problem Expressivity of Mixture-of-Experts models with varying granularity.
method Comparing models with different numbers of active experts (granularity).
result Exponential separation in network expressivity based on granularity.
We consider high-dimensional distribution estimation through autoregressive networks. By combining the concepts of sparsity, mixtures and parameter sharing we obtain a simple model which is fast to train and which achieves state-of-the-art or better results on several standard benchmark datasets. Specifically, we use a…
In this paper, we propose a generalized scale mixture family of distributions, namely the Power Exponential Scale Mixture (PESM) family, to model the sparsity inducing priors currently in use for sparse signal recovery (SSR). We show that the successful and popular methods such as LASSO, Reweighted ℓ1 and Reweigh…
MixMin finds optimal data mixtures for better model performance.
problem Finding the best data mixtures for improved model performance is challenging.
method Developed a gradient-based approach for optimizing a convex bi-level objective.
result MixMin mixtures uniformly improved model performance across various tasks.