Laplace Approximation For Tensor Train Kernel Machines In System Identification

arXiv — stat.MLWednesday, December 3, 2025 at 5:00:00 AM
  • A new study introduces a Bayesian tensor train kernel machine that utilizes Laplace approximation to enhance Gaussian process regression in system identification. This method addresses scalability issues by estimating the posterior distribution over a selected tensor train core while employing variational inference for hyperparameter precision. Experiments indicate that core selection is largely independent of tensor train ranks and feature structures, achieving significantly faster training times.
  • The development of this Bayesian tensor train kernel machine is significant as it offers a robust solution to the computational challenges faced in Gaussian process regression, particularly in high-dimensional data scenarios. By replacing traditional cross-validation with variational inference, the method not only improves efficiency but also maintains accuracy, which is crucial for applications in dynamic systems and machine learning.
  • This advancement reflects a broader trend in artificial intelligence where researchers are increasingly focusing on improving the scalability and efficiency of probabilistic models. The integration of variational inference and tensor networks is becoming a common theme, as seen in various studies addressing complex Bayesian models and optimization techniques. These developments highlight the ongoing efforts to refine machine learning methodologies, particularly in handling large datasets and enhancing model interpretability.
— via World Pulse Now AI Editorial System

Was this article worth reading? Share it

Recommended apps based on your readingExplore all apps
Continue Readings
Multi-view Bayesian optimisation in an input-output reduced space for engineering design
PositiveArtificial Intelligence
A recent study introduces a multi-view Bayesian optimisation approach that enhances the efficiency of Gaussian process models in engineering design by identifying a low-dimensional latent subspace from input and output data. This method utilizes probabilistic partial least squares (PPLS) to improve the scalability of Bayesian optimisation techniques in complex design scenarios.
Fast Gaussian Process Approximations for Autocorrelated Data
PositiveArtificial Intelligence
A new paper has been published addressing the computational challenges of Gaussian process models when applied to autocorrelated data, highlighting the risk of temporal overfitting if autocorrelation is ignored. The authors propose modifications to existing fast Gaussian process approximations to work effectively with blocked data, which helps mitigate these issues.
Adaptive Pruning for Increased Robustness and Reduced Computational Overhead in Gaussian Process Accelerated Saddle Point Searches
PositiveArtificial Intelligence
Recent advancements in Gaussian process (GP) regression have led to the development of an adaptive pruning strategy aimed at enhancing the robustness and efficiency of saddle point searches in high-dimensional energy landscapes. This approach utilizes geometry-aware optimal transport measures and a permutation-invariant metric to optimize computational overhead and improve stability during hyperparameter optimization.