5.0
Kernel Density Matrices (KDMs) provide a differentiable, compositional, and reversible way to represent joint probability distributions for both continuous and discrete variables
We adapt a foundation model SDO-FM originally trained on solar imagery to produce embeddings suitable for solar wind structure analysis.
We present a scalable machine learning framework for analyzing **Parker Solar Probe (PSP)** solar wind data using distributed processing and the quantum-inspired Kernel Density Matrices (KDM) method.
The **Clustering and Indexation Pipeline with Human Evaluation for Recognition (CIPHER)** framework combines interpretable compression (iSAX), density‑based clustering (HDBSCAN), and a human‑in‑the‑loop evaluation step to generate large‑scale, expert‑validated labels.
A key outcome is the use of the *magnetic deflection angle (θB)* as a unifying diagnostic across phenomena. The resulting system provides a scalable, interpretable, expert‑validated approach to solar wind classification, producing expanded event catalogs and aiding improved space weather forecasting.
INQMAD, a novel method for streaming anomaly detection that uses incremental density estimation with random Fourier features, quantum measurement principles, and density matrices to address these challenges.
We present the first large scale benchmark of 48 deep learning optical flow models on RADARSAT 2 ScanSAR sea ice imagery, evaluated with endpoint error (EPE) and Fl all metrics against GNSS tracked buoys.
M3LEO addresses the gap by providing a multi‑modal, multi‑label EO dataset that integrates polarimetric, interferometric, and coherence SAR data from Sentinel‑1 with multispectral imagery from *Sentinel‑2* and auxiliary terrain descriptors (e.g., land use).
Our research focuses on evaluating multiple tabular machine-learning models using the height information derived from the tomographic image intensities to classify eight distinct tree species.
Our method attempts to bypass traditional tomographic signal processing, potentially reducing latency from SAR capture to end product.
We use the TomoSense dataset, containing SAR and LiDAR data from Germany's Eifel National Park, to develop and evaluate height estimation models. Our approach includes classical methods, deep learning with a 3D U-Net, and Bayesian-optimized techniques.
This paper introduces a novel anomaly detection framework that combines the robust statistical principles of density-estimation-based anomaly detection methods with the representation-learning capabilities of deep learning models.
We present a novel quantum-classical density matrix density estimation model, called Q-DEMDE, based on the expected values of density matrices and a novel quantum embedding called quantum Fourier features.
This paper introduces a self-supervised pretraining scheme based on masked autoencoding applied to SAR amplitude data covering 8.7 % of the Earth’s land surface.
This paper presents a novel density estimation method for anomaly detection using density matrices (a powerful mathematical formalism from quantum mechanics) and Fourier features.
The novel density kernel density estimation method (DMKDE) uses density matrices, a quantum mechanical formalism, and random Fourier features, an explicit kernel approximation, to produce density estimates.
This paper explores how density matrices can be used as a building block for machine learning models exploiting their ability to straightforwardly combine linear algebra and probability.
We proposed a prediction model based on a machine learning approach. The method was trained with the random forest algorithm with historical data provided by each SEE. Three consecutive periods of data were concatenated.
This paper presents an anomaly detection model that combines the strong statistical foundation of density-estimation-based anomaly detection methods with the representation-learning ability of deep-learning models.
