PMODE: Theoretically Grounded and Modular Mixture Modeling
By: Robert A. Vandermeulen
Potential Business Impact:
Helps computers find weird data in huge amounts.
We introduce PMODE (Partitioned Mixture Of Density Estimators), a general and modular framework for mixture modeling with both parametric and nonparametric components. PMODE builds mixtures by partitioning the data and fitting separate estimators to each subset. It attains near-optimal rates for this estimator class and remains valid even when the mixture components come from different distribution families. As an application, we develop MV-PMODE, which scales a previously theoretical approach to high-dimensional density estimation to settings with thousands of dimensions. Despite its simplicity, it performs competitively against deep baselines on CIFAR-10 anomaly detection.
Similar Papers
MODE: Learning compositional representations of complex systems with Mixtures Of Dynamical Experts
Machine Learning (CS)
Helps predict how cells change and grow.
Mode-Conditioning Unlocks Superior Test-Time Scaling
Machine Learning (CS)
Makes AI smarter by exploring more ideas.
MLPMoE: Zero-Shot Architectural Metamorphosis of Dense LLM MLPs into Static Mixture-of-Experts
Machine Learning (CS)
Makes AI smarter and faster without retraining.