The recent developments in the research area demonstrate a strong trend towards integrating deep learning with traditional computational methods to address complex, high-dimensional problems across various domains. A significant focus is on leveraging generative models, such as Variational Autoencoders (VAEs) and Generative Adversarial Networks (GANs), to enhance the efficiency and accuracy of simulations and inference processes. These models are being employed to solve inverse problems, downscale weather forecasts, and model physical systems with greater interpretability and precision. Additionally, there is a notable shift towards incorporating quantum computing and physics-informed constraints into these models, which promises to revolutionize computational capabilities in fields like particle physics and climate modeling. The integration of active learning schemes and Bayesian inference methods is also advancing sample-efficient simulation-based inference, making it more feasible for high-dimensional settings. Overall, the field is progressing towards more hybrid, scalable, and interpretable solutions that bridge the gap between traditional methods and modern machine learning techniques.
Integrating Deep Learning and Computational Methods for Complex Problem Solving
Sources
Solving High-dimensional Inverse Problems Using Amortized Likelihood-free Inference with Noisy and Incomplete Data
Zephyr quantum-assisted hierarchical Calo4pQVAE for particle-calorimeter interactions
DPGIIL: Dirichlet Process-Deep Generative Model-Integrated Incremental Learning for Clustering in Transmissibility-based Online Structural Anomaly Detection
Using Machine Learning to Discover Parsimonious and Physically-Interpretable Representations of Catchment-Scale Rainfall-Runoff Dynamics
Variational Encoder-Decoders for Learning Latent Representations of Physical Systems
Active Sequential Posterior Estimation for Sample-Efficient Simulation-Based Inference
A Physics-Constrained Neural Differential Equation Framework for Data-Driven Snowpack Simulation
Enhancing operational wind downscaling capabilities over Canada: Application of a Conditional Wasserstein GAN methodology
GenAI4UQ: A Software for Inverse Uncertainty Quantification Using Conditional Generative Models
Sampling from Boltzmann densities with physics informed low-rank formats
Statistical Downscaling via High-Dimensional Distribution Matching with Generative Models
Disentangling impact of capacity, objective, batchsize, estimators, and step-size on flow VI
Deep clustering using adversarial net based clustering loss
Deep Clustering using Dirichlet Process Gaussian Mixture and Alpha Jensen-Shannon Divergence Clustering Loss
Stochastic Learning of Non-Conjugate Variational Posterior for Image Classification