Decision forests, including Random Forests and Gradient Boosting Trees, have recently demonstrated state-of-the-art performance in a variety of machine learning settings. Decision forests are typically ensembles of axis-aligned decision trees; that i
s, trees that split only along feature dimensions. In contrast, many recent extensions to decision forests are based on axis-oblique splits. Unfortunately, these extensions forfeit one or more of the favorable properties of decision forests based on axis-aligned splits, such as robustness to many noise dimensions, interpretability, or computational efficiency. We introduce yet another decision forest, called Sparse Projection Oblique Randomer Forests (SPORF). SPORF uses very sparse random projections, i.e., linear combinations of a small subset of features. SPORF significantly improves accuracy over existing state-of-the-art algorithms on a standard benchmark suite for classification with >100 problems of varying dimension, sample size, and number of classes. To illustrate how SPORF addresses the limitations of both axis-aligned and existing oblique decision forest methods, we conduct extensive simulated experiments. SPORF typically yields improved performance over existing decision forests, while mitigating computational efficiency and scalability and maintaining interpretability. SPORF can easily be incorporated into other ensemble methods such as boosting to obtain potentially similar gains.
Superconductors with persistent zero-resistance currents serve as permanent magnets for high-field applications requiring a strong and stable magnetic field, such as magnetic resonance imaging (MRI). The recent global helium shortage has quickened re
search into high-temperature superconductors (HTSs) materials that can be used without conventional liquid-helium cooling to 4.2 K. Herein, we demonstrate that 40-K-class metallic HTS magnesium diboride (MgB2) makes an excellent permanent bulk magnet, maintaining 3 T at 20 K for 1 week with an extremely high stability (<0.1 ppm/h). The magnetic field trapped in this magnet is uniformly distributed, as for single-crystalline neodymium-iron-boron. Magnetic hysteresis loop of the MgB2 permanent bulk magnet was detrmined. Because MgB2 is a simple-binary-line compound that does not contain rare-earth metals, polycrystalline bulk material can be industrially fabricated at low cost and with high yield to serve as strong magnets that are compatible with conventional compact cryocoolers, making MgB2 bulks promising for the next generation of Tesla-class permanent-magnet applications.
We have developed disk-shaped MgB2 bulk superconducting magnets (20, 30 mm in diameter, 10 mm in thickness) using the in-situ process from Mg and B powders and evaluated the temperature dependence of trapped magnetic field. A pair of two disc-shaped
bulks of 30 mm in diameter and 10 mm in thickness magnetized by field-cooling condition showed trapped fields of 1.2, 2.8 and 3.1 T at 30, 20 and 17.5 K, respectively. High trapped field over 3 T was recorded for the first time.
