Band selection is a common approach to reduce the data dimensionality of hyperspectral imagery. It extracts several bands of importance in some sense by taking advantage of high spectral correlation. ...Driven by detection or classification accuracy, one would expect that, using a subset of original bands, the accuracy is unchanged or tolerably degraded, whereas computational burden is significantly relaxed. When the desired object information is known, this task can be achieved by finding the bands that contain the most information about these objects. When the desired object information is unknown, i.e., unsupervised band selection, the objective is to select the most distinctive and informative bands. It is expected that these bands can provide an overall satisfactory detection and classification performance. In this letter, we propose unsupervised band selection algorithms based on band similarity measurement. The experimental result shows that our approach can yield a better result in terms of information conservation and class separability than other widely used techniques.
We propose a paradigm to apply machine learning various databases which have emerged in the study of the string landscape. In particular, we establish neural networks as both classifiers and ...predictors and train them with a host of available data ranging from Calabi–Yau manifolds and vector bundles, to quiver representations for gauge theories, using a novel framework of recasting geometrical and physical data as pixelated images. We find that even a relatively simple neural network can learn many significant quantities to astounding accuracy in a matter of minutes and can also predict hithertofore unencountered results, whereby rendering the paradigm a valuable tool in physics as well as pure mathematics.
It can be a great challenge for second language (L2) learners to comprehend meanings that are implied in utterances rather than the surface meaning of what was said. Moreover, L2 learners' attitudes ...toward pragmatic learning are unknown. This mixed-methods study investigates L2 learners' ability to comprehend conversational implicatures. It also explores their beliefs about and intentions to develop this ability using Ajzen's theory of planned behavior (TPB). A total of 498 freshmen from a public university in China participated in the study. Data were collected using a web-based test, stimulated recall tasks and semi-structured interviews. Results show that the participants differed in recognizing the intended meanings. Complicated factors account for the variations. In addition to the types of implicature, learners' beliefs about developing pragmatic comprehension also influence their learning intention, and subsequent performance. These beliefs include learners' multi-layered, complex attitudes toward the outcomes of pragmatic learning, perceived self-efficacy beliefs regarding language proficiency and L2 cultural knowledge, actual behavioral control over opportunities and resources for pragmatic learning, and perceptions of less social pressure on pragmatic learning. The use of TPB may help language teachers and test designers to understand learners' beliefs about L2 pragmatic learning in the English as a foreign language (EFL) context. Understanding the factors influencing learners' intention will help design more effective teaching curricula that may integrate pragmatic instruction and testing in the future.
Over the past decades, interest in shape memory polymers (SMPs) has persisted, and immense efforts have been dedicated to developing SMPs and their multifunctional composites. As a class of ...stimuli‐responsive polymers, SMPs can return to their initial shape from a programmed temporary shape under external stimuli, such as light, heat, magnetism, and electricity. The introduction of functional materials and nanostructures results in shape memory polymer composites (SMPCs) with large recoverable deformation, enhanced mechanical properties, and controllable remote actuation. Because of these unique features, SMPCs have a broad application prospect in many fields covering aerospace engineering, biomedical devices, flexible electronics, soft robotics, shape memory arrays, and 4D printing. Herein, a comprehensive analysis of the shape recovery mechanisms, multifunctionality, applications, and recent advances in SMPs and SMPCs is presented. Specifically, the combination of functional, reversible, multiple, and controllable shape recovery processes is discussed. Further, established products from such materials are highlighted. Finally, potential directions for the future advancement of SMPs are proposed.
Shape memory polymers are promising stimuli‐responsive materials that have provoked the interest of scientists over the past several decades. Progress in the mechanisms and applications of shape memory polymers is summarized, with a focus on the design and regulation of the shape memory effect. Future prospects and challenges in this exciting field are also proposed.
Extreme hypoxia of tumors represents the most notable barrier against the advance of tumor treatments. Inspired by the biological nature of red blood cells (RBCs) as the primary oxygen supplier in ...mammals, an aggressive man‐made RBC (AmmRBC) is created to combat the hypoxia‐mediated resistance of tumors to photodynamic therapy (PDT). Specifically, the complex formed between hemoglobin and enzyme‐mimicking polydopamine, and polydopamine‐carried photosensitizer is encapsulated inside the biovesicle that is engineered from the recombined RBC membranes. The mean corpuscular hemoglobin of AmmRBCs reaches about tenfold as high as that of natural RBCs. Owing to the same origin of outer membranes, AmmRBCs share excellent biocompatibility with parent RBCs. The introduced polydopamine plays the role of the antioxidative enzymes existing inside RBCs to effectively prevent the oxygen‐carrying hemoglobin from the oxidation damage during the circulation. This biomimetic engineering can accumulate in tumors, permit in situ efficient oxygen supply, and impose strong PDT efficacy toward the extremely hypoxic tumor with complete tumor elimination. The man‐made pseudo‐RBC shows potentials as a universal oxygen‐self‐supplied platform to sensitize hypoxia‐limited tumor treatment means, including but not limited to PDT. Meanwhile, this study offers ideas to the production of artificial substitutes of packed RBCs for clinical blood transfusion.
Aggressive man‐made pseudo‐red blood cells (AmmRBCs) are prepared for self‐oxygen‐supplied photodynamic therapy (PDT) toward tumors. AmmRBCs can accumulate in tumors and exhibit high efficacy to combat hypoxia‐induced resistance of tumors to PDT, leading to complete tumor elimination.
Previous works utilized "smaller-norm-less-important" criterion to prune filters with smaller norm values in a convolutional neural network. In this paper, we analyze this norm-based criterion and ...point out that its effectiveness depends on two requirements that are not always met: (1) the norm deviation of the filters should be large; (2) the minimum norm of the filters should be small. To solve this problem, we propose a novel filter pruning method, namely Filter Pruning via Geometric Median (FPGM), to compress the model regardless of those two requirements. Unlike previous methods, FPGM compresses CNN models by pruning filters with redundancy, rather than those with"relatively less" importance. When applied to two image classification benchmarks, our method validates its usefulness and strengths. Notably, on CIFAR-10, FPGM reduces more than 52% FLOPs on ResNet-110 with even 2.69% relative accuracy improvement. Moreover, on ILSVRC-2012, FPGM reduces more than 42% FLOPs on ResNet-101 without top-5 accuracy drop, which has advanced the state-of-the-art. Code is publicly available on GitHub: https://github.com/he-y/filter-pruning-geometric-median.
The metabolic switch from oxidative phosphorylation to glycolysis is required for tumorigenesis in order to provide cancer cells with energy and substrates of biosynthesis. Therefore, it is important ...to elucidate mechanisms controlling the cancer metabolic switch. MTR4 is a RNA helicase associated with a nuclear exosome that plays key roles in RNA processing and surveillance. We demonstrate that MTR4 is frequently overexpressed in hepatocellular carcinoma (HCC) and is an independent diagnostic marker predicting the poor prognosis of HCC patients. MTR4 drives cancer metabolism by ensuring correct alternative splicing of pre-mRNAs of critical glycolytic genes such as GLUT1 and PKM2. c-Myc binds to the promoter of the MTR4 gene and is important for MTR4 expression in HCC cells, indicating that MTR4 is a mediator of the functions of c-Myc in cancer metabolism. These findings reveal important roles of MTR4 in the cancer metabolic switch and present MTR4 as a promising therapeutic target for treating HCC.
Machine Learning Calabi–Yau Metrics Ashmore, Anthony; He, Yang‐Hui; Ovrut, Burt A.
Fortschritte der Physik,
September 2020, Volume:
68, Issue:
9
Journal Article
Peer reviewed
Open access
We apply machine learning to the problem of finding numerical Calabi–Yau metrics. Building on Donaldson's algorithm for calculating balanced metrics on Kähler manifolds, we combine conventional curve ...fitting and machine‐learning techniques to numerically approximate Ricci‐flat metrics. We show that machine learning is able to predict the Calabi–Yau metric and quantities associated with it, such as its determinant, having seen only a small sample of training data. Using this in conjunction with a straightforward curve fitting routine, we demonstrate that it is possible to find highly accurate numerical metrics much more quickly than by using Donaldson's algorithm alone, with our new machine‐learning algorithm decreasing the time required by between one and two orders of magnitude.
The concept of machine‐learning is applied to the problem of finding numerical Calabi–Yau metrics. Building on Donaldson's algorithm for calculating balanced metrics on Kähler manifolds, conventional curve fitting and machine‐learning techniques are combined to numerically approximate Ricci‐flat metrics. It is shown that machine learning is able to predict the Calabi–Yau metric and quantities associated with it, such as its determinant, having seen only a small sample of training data. Using this in conjunction with a straightforward curve fitting routine, the authors demonstrate that it is possible to find highly accurate numerical metrics much more quickly than by using Donaldson's algorithm alone, with the new machine‐learning algorithm decreasing the time required by between one and two orders of magnitude.
A new, complete sample of 14,584 broad-line active galactic nuclei (AGNs) at z < 0.35 is presented, which are uncovered homogeneously from the complete database of galaxies and quasars observed ...spectroscopically in the Sloan Digital Sky Survey Seventh Data Release. The stellar continuum is properly removed for each spectrum with significant host absorption line features, and careful analyses of the emission line spectra, particularly in the H and Hβ wavebands, are carried out. The broad Balmer emission line, particularly H , is used to indicate the presence of an AGN. The broad H lines have luminosities in a range of 1038.5- 1044.3 , and line widths (FWHMs) of 500-34,000 . The virial black hole masses, estimated from the broad-line measurements, span a range of 105.1- 1010.3 , and the Eddington ratios vary from −3.3 to 1.3 in logarithmic scale. Other quantities such as multiwavelength photometric properties and flags denoting peculiar line profiles are also included in this catalog. We describe the construction of this catalog and briefly discuss its properties. The catalog is publicly available online. This homogeneously selected AGN catalog, along with the accurately measured spectral parameters, provides the most updated, largest AGN sample data, which will enable further comprehensive investigations of the properties of the AGN population in the low-redshift universe.