Abstract: Provided are improved end-to-end self-supervised pre-training frameworks that leverage a combination of contrastive and masked modeling loss terms. In particular, the present disclosure provides framework that combines contrastive learning and masked modeling, where the former trains the model to discretize input data (e.g., continuous signals such as continuous speech signals) into a finite set of discriminative tokens, and the latter trains the model to learn contextualized representations via solving a masked prediction task consuming the discretized tokens. In contrast to certain existing masked modeling-based pre-training frameworks which rely on an iterative re-clustering and re-training process or other existing frameworks which concatenate two separately trained modules, the proposed framework can enable a model to be optimized in an end-to-end fashion by solving the two self-supervised tasks (the contrastive task and masked modeling) simultaneously.

Abstract: The present disclosure provide a detection device of microfluidic polymerase chain reaction (PCR) and a detection system including the same. This all-in-one device and system may be used to detect at least one biological detection chip, so that can amplify gene fragments at the front-end and detect them at the back-end immediately, decreasing the time required for the analysis, enabling real-time, low-cost, and rapid detection of various viruses, such as EBV and COVID-19, without compromising accuracy or sensitivity.

Abstract: An attached floating image touch-control device is provided. The attached floating image touch-control device includes a trigger module, a floating image generation module and a sensing module. The trigger module includes a movable trigger component. The floating image generation module includes a light source, an imaging unit, and a floating imaging unit disposed on the imaging unit. The sensing module is arranged on the floating image generation module and communicatively connected to the trigger module, wherein when the sensing module senses a touch action, the sensing module outputs a driving signal, and the trigger module moves the trigger component based on the driving signal to contact and trigger the contact-type button.

Abstract: A pixel circuit, a driving method therefor and a display apparatus are provided. The pixel circuit includes a driving sub-circuit, a write sub-circuit, a compensation sub-circuit, a reset sub-circuit, a first light-emitting control sub-circuit, a second light-emitting control sub-circuit, a leak-proof sub-circuit, a storage sub-circuit and a light-emitting element. The reset sub-circuit is configured to reset a fourth node under control of a signal of a light-emitting control signal terminal and reset a fifth node under control of a signal of a reset control signal terminal. The compensation sub-circuit is configured to compensate a threshold voltage of the driving sub-circuit to the fifth node under the control of a signal of a first scanning signal terminal. The leak-proof sub-circuit is configured to write a signal of the fifth node into a first node under control of a signal of a second scanning signal terminal.

Abstract: A method of reducing a concentration of a nitrogen oxide, the method comprising: contacting a microorganism with a nitrogen oxide-containing sample to reduce the concentration of the nitrogen oxide in the sample, wherein the contacting comprises contacting the microorganism with Fe(II)(L)-NOx in a bioreactor, wherein the Fe(II)(L)-NOx is a complex in which a chelating agent, Fe2+, and NOx are chelated, wherein L is the chelating agent, and wherein NOx is a nitrogen oxide ligand.

Abstract: A device includes a sensor die having a sensing region at a top surface of the sensor die, an encapsulant at least laterally encapsulating the sensor die, a conductive via extending through the encapsulant, and a front-side redistribution structure on the encapsulant and on the top surface of the sensor die, wherein the front-side redistribution structure is connected to the conductive via and the sensor die, wherein an opening in the front-side redistribution structure exposes the sensing region of the sensor die, and wherein the front-side redistribution structure includes a first dielectric layer extending over the encapsulant and the top surface of the sensor die, a metallization pattern on the first dielectric layer, and a second dielectric layer extending over the metallization pattern and the first dielectric layer.

Abstract: A method of decoding a bitstream by an electronic device is provided. A block unit is determined from an image frame received from the bitstream. An intra prediction mode index corresponding to one of wide-angle candidate modes is determined for the block unit. The electronic device determines whether the intra prediction mode index is different from predefined indices each corresponding to one of predefined wide-angle modes in the wide-angle candidate modes. Filtered samples are generated based on reference samples neighboring the block unit. The filtered samples are generated by an interpolation filter when the intra prediction mode index is different from the predefined indices. The filtered samples are generated by a reference filter when the intra prediction mode index is equal to at least one of the predefined indices. The block unit is reconstructed based on the filtered samples along a mode direction of the intra prediction mode index.

Abstract: In an embodiment, a device includes: a molding compound; an integrated circuit die encapsulated in the molding compound; a through via adjacent the integrated circuit die; and a redistribution structure over the integrated circuit die, the molding compound, and the through via, the redistribution structure electrically connected to the integrated circuit die and the through via, the redistribution structure including: a first dielectric layer disposed over the molding compound; a first conductive via extending through the first dielectric layer; a second dielectric layer disposed over the first dielectric layer and the first conductive via; and a second conductive via extending through the second dielectric layer and into a portion of the first conductive via, an interface between the first conductive via and the second conductive via being non-planar.

Abstract: The present disclosure describes a method of forming low thermal budget dielectrics in semiconductor devices. The method includes forming, on a substrate, first and second fin structures with an opening in between, filling the opening with a flowable isolation material, treating the flowable isolation material with a plasma, and removing a portion of the plasma-treated flowable isolation material between the first and second fin structures.

Abstract: A transparent display device is provided. The transparent display device includes a display unit having a circuit area and a transparent area. The display unit includes a plurality of signal lines located in the circuit area, a plurality of pixel circuits electrically connected to the signal lines and located in the circuit area, a plurality of light-emitting elements driven by the pixel circuits and located in the circuit area, and an encapsulation layer located in the circuit area and the transparent area. A first thickness of the encapsulation layer located in the circuit area is different from a second thickness of the encapsulation layer located in the transparent area.

Abstract: A device includes a first channel layer, a second channel layer, a gate structure, a source/drain epitaxial structure, and a source/drain contact. The first channel layer and the second channel layer are arranged above the first channel layer in a spaced apart manner over a substrate. The gate structure surrounds the first and second channel layers. The source/drain epitaxial structure is connected to the first and second channel layers. The source/drain contact is connected to the source/drain epitaxial structure. The second channel layer is closer to the source/drain contact than the first channel layer is to the source/drain contact, and the first channel layer is thicker than the second channel layer.

Abstract: An HOD device, comprising: a framework; covering material, covering the frame work; at least one conductive region, provided on or in the covering material; wherein the conductive region is coupled to a capacitance detection circuit or a predetermined voltage level. The HOD device can be a vehicle control device such as a steering wheel. The conductive region comprises conductive wires which can be threads of the covering material. By this way, the arrangements of the conductive wires can be changed corresponding to the size or the shape of the frame work or any other requirements. Also, the interference caused by unstable factors can be improved since the conductive wires can be coupled to a ground source of the vehicle to provide a short capacitance sensing path.

Abstract: Methods, systems, and devices are disclosed for imaging particles and/or cells using flow cytometry. In one aspect, a method includes transmitting a light beam at a fluidic channel carrying a fluid sample containing particles; optically encoding scattered or fluorescently-emitted light at a spatial optical filter, the spatial optical filter including a surface having a plurality of apertures arranged in a pattern along a transverse direction opposite to particle flow and a longitudinal direction parallel to particle flow, such that different portions of a particle flowing over the pattern of the apertures pass different apertures at different times and scatter the light beam or emit fluorescent light at locations associated with the apertures; and producing image data associated with the particle flowing through the fluidic channel based on the encoded optical signal, in which the produced image data includes information of a physical characteristic of the particle.

Abstract: A hybrid method for green intelligent manufacturing (GiM) combines the carbon reduction and the energy saving into the intelligent manufacturing based Industry 4.1 cloud platform. GiM assists companies to achieve the goal of net zero transition and help them advance to Industry 4.2 as soon as possible by simultaneously taking carbon footprint and energy issues into account. GiM collects large volumes of essential data (including carbon footprint) via cyber physical agents (CPAs), and sends them to two critical services of carbon management and intelligent energy management system (iEMS) deployed on the cloud platform. The two critical services optimize the energy dispatch schedule by strictly following the requirements of energy saving, carbon reduction, and net zero. Then, the state of zero defects of intelligent manufacturing achieved in Industry 4.1 can be upgraded to net zero of GiM in Industry 4.2.

Abstract: A mask module includes a framework and a first strip plate fixed on the framework and extending along a first direction. The first strip plate is provided with a first concave-convex structure on one side edge along the first direction. At least one convex structure is provided in a middle of the first concave-convex structure along the first direction. The convex structure and the first stripe plate are integrally formed.

Abstract: A mask module includes a framework, a first strip plate fixed on the framework and extending along a first direction, and a first mask. The first mask is located on a side, deviating from the framework, of the first strip plate. The first mask includes at least one preset area, and the preset area includes at least one opening area. The first strip plate is provided with a first concave-convex structure on one side edge along the first direction. At least one convex structure is provided in a middle of the first concave-convex structure along the first direction. In a direction perpendicular to a surface of the first mask, the first concave-convex structure and the convex structure cover at least a part of area of the at least one opening of the preset area. The convex structure and the first stripe plate are integrally formed.

Abstract: Embodiments of the present invention provide a hybrid system and method for distributed virtual power plants integrated intelligent net zero. In this method, a cyber physical agent (CPA) is utilized to collect a carbon emission information and an energy management information, and then an artificial intelligence (AI) optimization model of an intelligent central dispatch platform is utilized to obtain a power dispatch manner of the distributed virtual power plants based on the carbon emission information and the energy management information, such that the power dispatch manner of the distributed virtual power plants meets the requirements of enterprise economic benefits and net zero carbon emissions at the same time.

Abstract: A flexible permanent magnetic material, a preparation method and an application thereof in magnetic biological effect products are provides, relating to the technical field of medical equipment. Raw materials of the flexible permanent magnetic material of the application include the following components in parts by weight: 0-70 parts of anisotropic neodymium iron boron powder and 0-40 parts of anisotropic samarium iron nitrogen powder and 3-20 parts of binder.

Abstract: A memory device includes at least one bit line, at least one word line, at least one memory cell, at least one source line, and a controller electrically coupled to the at least one memory cell via the at least one word line, the at least one bit line, and the at least one source line. The memory cell includes a first transistor, data storage elements, and second transistors corresponding to the data storage elements. The first transistor includes a gate electrically coupled to the word line, and first and second source/drains. Each data storage element and the corresponding second transistor are electrically coupled in series with the first source/drain of the first transistor and the bit line. The controller controllably applies a voltage other than a ground voltage to the at least one source line in an operation of a selected data storage element among the data storage elements.

Abstract: The described aspects and implementations enable efficient detection and classification of objects with machine learning models that deploy a bird's-eye view representation and are trained using depth ground truth data. In one implementation, disclosed are system and techniques that include obtaining images, generating, using a first neural network (NN), feature vectors (FVs) and depth distributions pixels of images, wherein the first NN is trained using training images and a depth ground truth data for the training images. The techniques further include obtaining a feature tensor (FT) in view of the FVs and the depth distributions, and processing the obtained FTs, using a second NN, to identify one or more objects depicted in the images.