Abstract: Methods, apparatus, systems, and articles of manufacture are disclosed for teacher-free self-feature distillation training of machine-learning (ML) models. An example apparatus includes at least one memory, instructions, and processor circuitry to at least one of execute or instantiate the instructions to perform a first comparison of (i) a first group of a first set of feature channels (FCs) of an ML model and (ii) a second group of the first set, perform a second comparison of (iii) a first group of a second set of FCs of the ML model and one of (iv) a third group of the first set or a first group of a third set of FCs of the ML model, adjust parameter(s) of the ML model based on the first and/or second comparisons, and, in response to an error value satisfying a threshold, deploy the ML model to execute a workload based on the parameter(s).

Abstract: The invention discloses a method for synthesizing 1,7-naphthyridine derivatives, which relates to the technical field of synthesizing pharmaceutical intermediates and organic chemical intermediates, wherein the method includes: (1) 2-chloro-3-amino-pyridine being used as Compound I as a starting material, and protecting an amino group to prepare Compound II; (2) the Compound II reacting with an aldehydation reagent under alkaline conditions to obtain Compound III; (3) cyclizing the Compound III with acrylate compounds under the action of Lewis acid to prepare compound IV.

Abstract: A gene related to biosynthesis of ergothioneine is provided. Based on five synthesis genes of ergothioneine in Mycobacteroides abscessus, only an encoding sequence of each of the five genes is retained, and a gene structure of an encoding region in each gene is optimized to obtain genes related to the biosynthesis of the ergothioneine that are stably expressed in rice. These genes have sequences set forth in SEQ ID NO: 1 to SEQ ID NO: 5, and encoded proteins thereof have amino acid sequences set forth in SEQ ID NO: 6 to SEQ ID NO: 10. The gene is transferred into rice to construct a transgenic rice molecular farm to produce the ergothioneine. As a molecular farm, rice seeds do not contain alkaloids or allergens that are harmful to the human body. Moreover, biosynthesized ergothioneine shows the advantages of easy extraction and processing and of low production cost.

Abstract: Systems, methods and devices for a computer vision-based pixel-level rail components detection system using an improved one-stage instance segmentation model and prior knowledge, aiming to inspect railway components in a rapid, accurate, and convenient fashion.

Abstract: A mask instrument includes a flexible mask and a controller. The flexible mask includes a first flexible layer, a second flexible layer, a plurality of functional layers located between the first flexible layer and the second flexible layer, and a plurality of electrodes electrically connected with the plurality of functional layers. The flexible mask is electrically coupled with the controller via the plurality of electrodes. The second flexible layer comprises at least one hole at a position corresponding to the plurality functional layers, and at least one portion of the plurality of functional layer is exposed out of the second flexible layer from the at least one hole. A method for using the mask instrument is further provided.

Abstract: An array substrate has sub-pixel regions arranged in an array, and the sub-pixel regions include white sub-pixel regions and primary color sub-pixel regions. The array substrate includes a first substrate and a plurality of sub-pixels disposed on the first substrate. The sub-pixels include white sub-pixels and primary color sub-pixels. In a column direction, a side of each white sub-pixel is adjacent to at least one primary color sub-pixel. Each sub-pixel has light-shielding patterns. In the column direction, in a plurality of light-shielding patterns of a primary color sub-pixel adjacent to the white sub-pixel, a part of the light-shielding patterns are disposed in a white sub-pixel region where the white sub-pixel is located, and another part of the light-shielding patterns are disposed in a primary color sub-pixel region corresponding to the primary color sub-pixel.

Abstract: A system for predicting disease with graph convolutional neural network based on multimodal magnetic resonance imaging, which extracts the radiomics information of multiple brain regions across modals as the features of nodes from multimodal magnetic resonance data, and extracts the connectomics information between brain regions to form an adjacency matrix. T1-weighted structural images extract cortical information through cortical reconstruction, and resting-state magnetic resonance data are used to calculate amplitude of low frequency fluctuations, regional homogeneity and functional connectivity. Through multimodal data preprocessing, image index extraction and structured data integration, multimodal unstructured magnetic resonance image data are integrated into unified graph-structured data, and the disease is predicted by a graph convolutional neural network method.

Abstract: A system obtains a first training data set comprising labeled speech data or both labeled and unlabeled data corresponding to a high-resource data set as well as latent speech representations based on the first training data set. The system trains a machine learning model on the first training data set to learn phonetically aware speech representations corresponding to the first training data set. The system applies the latent speech representations to a transformer context network to generate contextual representations. The system aligns each of the contextual representations with a phoneme label to generate phonetically-aware contextual representations. The system causes a refinement engine to further refine the machine learning model.

Abstract: The present application discloses a power-on-reset circuit, which optimizes a hysteresis circuit and a reset signal generation circuit, and introduces a seventh PMOS transistor as a switch transistor to achieve the differentiation of control voltages at a gate end of a first NMOS transistor during powering-on and off. A voltage rise detection point is determined by a partial voltage of a resistor during powering-on, while a voltage fall detection point is directly determined by a power supply voltage during powering-off. Such differentiation may achieve a significant separation between the voltage rise detection point and the voltage fall detection point, reducing the voltage fall detection point to near a threshold voltage of the first NMOS transistor, and meeting the demand for a lower voltage fall detection point, which is consistent with a practical application of the power-on-reset circuit in an MCU.

Abstract: Described herein are a low-cost, non-destructive, and contact-free intelligent inspection system that is field-deployable on a geometry car, high-rail vehicle, or other types of track inspection platforms to identify broken railway/railroad spikes in real-time.

Abstract: The present disclosure provides, among other things, a method for efficiently producing a cell population enriched in Natural Killer cells (NK cells) from induced pluripotent cells.

Abstract: Evaluating railway ballast fouling condition is critical to assessing track conditions and arranging proper ballast maintenance. Because fouled ballast materials with different fouling conditions have different material properties, these properties can be used to evaluate the fouling severity. Various previously developed approaches to estimating fouling conditions often require special sensors or equipment, and well-trained technicians. Recently, convolutional neural network (CNN) based computer vision approaches have performed particle segmentation to obtain ballast grain size distribution. While the coarse aggregate fraction can be evaluated, many such approaches do not segment fine particles. This disclosure is an image analysis approach to directly estimate the ballast fouling conditions. First, fouled ballast images with different fouling conditions are taken as the reference. Then, the RGB color distributions of the fouled ballast images are processed through statistical analysis.

Abstract: The disclosure deals with a system and method recognizing the same person from different pictures in order to identify who is who based on photos or videos from different sources. Facial recognition plays an increasingly important role in modern life. Due to the low power consumption of Spiking Neural Networks (SNNs), this disclosure proposed an SNNs-based facial recognition model converted from a well-known facial recognition network ArcFace. The experimental results show that SNN (duration=200) achieved 99.00% overall accuracy on the testing dataset and outperformed the original ArcFace.

Abstract: Described herein are a low-cost, non-destructive, and contact-free intelligent inspection system that is field-deployable on a geometry car, high-rail vehicle, or other types of track inspection platforms to identify broken railway/railroad spikes in real-time.

Abstract: The present invention discloses an energy-efficient capacitance extraction method based on machine learning, and involves improving parameter extraction efficiency by using a machine learning model to extract parasitic capacitance; generally representing an interconnection line structure by grid-based data representation; reducing a workload of parameter extraction and enhancing the robustness of different semiconductor technologies with the idea of an adaptive extraction window; establishing a machine learning model of capacitance extraction for a two-dimensional interconnection line structure, and extracting grid parameters of a target interconnection line structure and inputting the grid parameters into the machine learning model, thereby obtaining parasitic capacitance parameters. Compared with an existing capacitance extraction technology, an capacitance extractor has achieved excellent performance in accuracy, speed and time and space consumption.

Abstract: Described herein are systems and methods for identifying an optimal travel route for first travelers, such as commuters, first responders, etc., that considers both potential railroad crossing blockages in addition to street traffic congestion conditions to provide real-time optimal vehicle routing based on real-time street and railroad crossing conditions.

Abstract: Described herein are systems, methods and devices for gathering sound emission data from railway spikes in order to determine if a spike is undamaged, damaged, or broken via initiating sound waves in the spike and analyzing same to determine the structural integrity of the spike being tested.

Abstract: Described herein is a fully autonomous drone-based track inspection system that does not rely on GPS but instead uses optical images taken from the drone to identify the railroad track and navigate the drone to cruise along the track to perform track inspection tasks; track images are taken by the onboard drone camera and processed to provide both navigation information for autonomous drone flight control and track component health evaluation.

Abstract: The present invention relates to a composite comprising (1) a bottom layer comprising expanded thermoplastic elastomer particles; and (2) a surface layer on the bottom layer, and the use of such composites in flooring surfaces sports, sports hall floorings, swimming pool hall floorings, running tracks, sports facilities, playgrounds, kindergartens, park walkway and pavements.

Abstract: Described herein are a low-cost, non-destructive, and contact-free intelligent inspection system that is field-deployable on a geometry car, high-rail vehicle, or other types of track inspection platforms to identify broken railway/railroad spikes in real-time.