Abstract: A method and system for estimating the ground state energy of a quantum Hamiltonian. The disclosed algorithm may run on any hardware and is suited for early fault tolerant quantum computers. The algorithm employs low-depth quantum circuits with one ancilla qubit with classical post-processing. The algorithm first draws samples from Hadamard tests in which the unitary is a controlled time evolution of the Hamiltonian. The samples are used for evaluating the convolution of the spectral measure and a filter function, and then inferring the ground state energy from this convolution. Quantum circuit depth is linear in the inverse spectral gap and poly-logarithmic in the inverse target accuracy and inverse initial overlap. Runtime is polynomial in the inverse spectral gap, inverse target accuracy, and inverse initial overlap. The algorithm produces a highly-accurate estimate of the ground state energy with reasonable runtime using low-depth quantum circuits.

Abstract: Systems, apparatuses and methods may provide technology for optimizing an inference neural network model that performs asymmetric quantization by generating a quantized neural network, wherein model weights of the neural network are quantized as signed integer values, and wherein an input layer of the neural network is configured to quantize input values as unsigned integer values, generating a weights accumulation table based on the quantized model weights and a kernel size for the neural network, and generating an output restoration function for an output layer of the neural network based on the weights accumulation table and the kernel size. The technology may also perform per-input channel quantization. The technology may also perform mixed-precision auto-tuning.

Abstract: The present disclosure discloses a machine learning-based method for defending a voice assistant from being controlled by an inaudible command, including following steps: 1) collecting data of positive and negative samples, 2) performing data segmentation on data of the positive and negative samples; 3) selecting and normalizing sample features; 4) selecting a classifier to be trained and generate a detection model for a malicious voice command; 5) detecting a voice command to be detected by the detection model. The present disclosure selects an original feature selection method, and for smart devices of different types, it is necessary to obtain normal voice commands and malicious voice commands by means of a smart device of this type, and use them as the positive and negative samples to train a specific classifier for the device. Such a customized approach can well solve a problem that detection and defense between devices cannot work.

Abstract: Provided in the present application is a manufacturing method for a flexible photovoltaic assembly comprising: 1) connecting cells in series to form a cell string; 2) connecting the cell strings in series and/or in parallel to form a cell layer, wherein a gap is present between adjacent cell strings; 3) sequentially stacking a front panel material layer, a front packaging material layer, the cell layer, a back packaging material layer, and a back panel material layer, and performing lamination to obtain a laminated member; 4) performing punching on the laminated member to remove the material layers that are located in the gap to obtain a photovoltaic assembly, the photovoltaic assembly comprising photovoltaic assembly units, and the photovoltaic assembly units are electrically connected by the electrical connection material; and 5) sequentially stacking a composite material layer, the photovoltaic assembly, a flexible substrate layer, performing lamination to obtain a flexible photovoltaic assembly.

Abstract: A building earthquake resistance structure includes an integrally connected roof support frame, used for supporting a roof of a building independently from walls of the building; a plurality of support columns, fixedly connected to the roof support frame; an annular trench, arranged in the ground around the building; an annular damping frame, arranged within the annular trench, lower ends of the support columns being fixedly connected to the annular damping frame; a plurality of dampers, arranged between a bottom portion of the annular trench and the annular damping frame. An earthquake resistance method for the building earthquake resistance structure prevents damage caused by the roof collapsing during an earthquake by supporting the roof of the existing building in the ground via the support columns. The building earthquake resistance structure is easy to construct and suitable for transforming traditional old buildings.

Abstract: A building earthquake resistance structure includes an integrally connected roof support frame, used for supporting a roof of a building independently from walls of the building; a plurality of support columns, fixedly connected to the roof support frame; an annular trench, arranged in the ground around the building; an annular damping came, arranged within the annular trench, lower ends of the support columns being fixedly connected to the annular damping frame; a plurality of dampers, arranged between a bottom portion of the annular trench and the annular damping frame. An earthquake resistance method for the building earthquake resistance structure prevents damage caused by the roof collapsing during an earthquake by supporting the roof of the existing building in the ground via the support columns. The building earthquake resistance structure is easy to construct and suitable for transforming traditional old buildings.

Abstract: A composite plate is mainly made from the following raw materials in percentage by weight: 35% to 45% of calcium carbonate, 45% to 55% of cullet, 5% to 15% of unsaturated polyester resin, 0.1% to 1% of a curing agent, 0.1% to 1% of an auxiliary agent, wherein a ratio of an amount of the calcium carbonate to an amount of the cullet is ?0.75%. The invention employs cullet, calcium carbonate, and resin as the main raw materials to prepare an artificial stone plate; no natural quartz is used, and hence the raw material cost is low. A reasonable ratio of cullet to calcium carbonate reduces the use of resin, which further reduces the raw material cost. Through a reasonable combination of different raw materials, the composite plate of the invention has physical and chemical properties and decorative abilities that are similar to those of traditional artificial quartz stone.

Abstract: A composite plate is mainly made from the following raw materials in percentage by weight: 35% to 45% of calcium carbonate, 45% to 55% of cullet, 5% to 15% of unsaturated polyester resin, 0.1% to 1% of a curing agent, 0.1% to 1% of an auxiliary agent, wherein a ratio of an amount of the calcium carbonate to an amount of the cullet is 0.75%. The invention employs cullet, calcium carbonate, and resin as the main raw materials to prepare an artificial stone plate; no natural quartz is used, and hence the raw material cost is low. A reasonable ratio of cullet to calcium carbonate reduces the use of resin, which further reduces the raw material cost. Through a reasonable combination of different raw materials, the composite plate of the invention has physical and chemical properties and decorative abilities that are similar to those of traditional artificial quartz stone.

Abstract: The present disclosure discloses a machine learning-based method for defending a voice assistant from being controlled by an inaudible command, including following steps: 1) collecting data of positive and negative samples, 2) performing data segmentation on data of the positive and negative samples; 3) selecting and normalizing sample features; 4) selecting a classifier to be trained and generate a detection model for a malicious voice command; 5) detecting a voice command to be detected by the detection model. The present disclosure selects an original feature selection method, and for smart devices of different types, it is necessary to obtain normal voice commands and malicious voice commands by means of a smart device of this type, and use them as the positive and negative samples to train a specific classifier for the device. Such a customized approach can well solve a problem that detection and defense between devices cannot work.

Abstract: The invention relates to the technical field of artificial stone surfaces, in particular to an artificial glass surface, which is made from the following raw materials in parts by mass: 0-30 parts of a quartz material, 40-70 parts of a glass material, 5-15 parts of a modified silicone resin, 8-15 parts of an unsaturated polyester resin, and 5-14 parts of additional raw materials. The artificial glass surface employs recycled glass material as its main stone source, which contributes the conservation of mineral resources, and reduces production costs; the product is of higher quality.

Abstract: The present invention provides a method for preparing an artificial glass surface, which includes the following steps: step A: mixing a glass material and a silane coupling agent to obtain an initial glass material mixture. Step B: heating a silicone resin, mixing with the initial glass material mixture thoroughly to obtain a secondary glass material wrapper. Step C: adding an unsaturated polyester resin, a curing agent, a filler pigment, and a quartz material to obtain a glass surface premade material. Step D: pressing the glass surface premade material, curing to obtain the artificial glass surface. In this invention, the quality of the product is maintained at a high level, mineral resources are conserved, and production cost is lowered.

Abstract: The present invention involves a multiplayer artificial stone and its preparation method. The decorative layer comprises the following components in percentage by weight: 80%-90% of a decorative stone, 5%-15% of an unsaturated polyester resin, 0.6%-2% of a coupling agent, 0.7%-1.2% of a curing agent, and 1%-5% of a pigment filler. The bottom layer comprises the following components in percentage by weight: 65%-78% of a calcium carbonate stone, 10%-18% of an unsaturated polyester resin, 0.6%-2% of a coupling agent, 0.7%-1.2% of a curing agent, and 1-15% of a pigment filler. The preparation method involves preparing the decorative layer and the bottom layer separately and then binding the two layers together by pressing. This arrangement allows the multilayer artificial stone to have good decorative effect and product quality; in addition, each layer contains only one type of stone, therefore its preparation is easier and can be more precisely controlled, lowering production cost.

Abstract: A back-contact cell module including cells and connecting ribbons; a main gate electrode for gathering currents of the auxiliary gate electrode and a linear back side electrode for leading out the currents provided on the back side of the cell; the main gate electrodes, located on the back side of the cell and in one-to-one correspondence with positions of current collection holes, are arranged linearly, parallel to the back side electrode and located at two ends of the back side of the cell; the plurality of cells are arranged parallelly in columns, between two adjacent cells, the main gate electrode of one cell is arranged adjacent to the back side electrode of the other cell; the main gate electrode of one cell is fixedly connected to the back side electrode of the other adjacent cell via the connecting ribbon.

Abstract: The invention relates to the technical field of artificial stone surfaces, in particular to an artificial glass surface, which is made from the following raw materials in parts by mass: 0-30 parts of a quartz material, 40-70 parts of a glass material, 5-15 parts of a modified silicone resin, 8-15 parts of an unsaturated polyester resin, and 5-14 parts of additional raw materials. The artificial glass surface employs recycled glass material as its main stone source, which contributes the conservation of mineral resources, and reduces production costs; the product is of higher quality.

Abstract: The present invention provides a method for preparing an artificial glass surface, which includes the following steps: step A: mixing a glass material and a silane coupling agent to obtain an initial glass material mixture. Step B: heating a silicone resin, mixing with the initial glass material mixture thoroughly to obtain a secondary glass material wrapper. Step C: adding an unsaturated polyester resin, a curing agent, a filler pigment, and a quartz material to obtain a glass surface premade material. Step D: pressing the glass surface premade material, curing to obtain the artificial glass surface. In this invention, the quality of the product is maintained at a high level, mineral resources are conserved, and production cost is lowered.

Abstract: A back-contact cell module including cells and connecting ribbons; a main gate electrode for gathering currents of the auxiliary gate electrode and a linear back side electrode for leading out the currents provided on the back side of the cell; the main gate electrodes, located on the back side of the cell and in one-to-one correspondence with positions of current collection holes, are arranged linearly, parallel to the back side electrode and located at two ends of the back side of the cell; the plurality of cells are arranged parallelly in columns, between two adjacent cells, the main gate electrode of one cell is arranged adjacent to the back side electrode of the other cell; the main gate electrode of one cell is fixedly connected to the back side electrode of the other adjacent cell via the connecting ribbon.

Abstract: The present disclosure provides a diesel fuel composition containing DMC and multifunctional additives to reduce particulate emission, improve efficiency and be used in cold and/or hypoxia conditions.

Abstract: The present disclosure provides a diesel fuel composition containing DMC and multifunctional additives to reduce particulate emission, improve efficiency and be used in cold and/or hypoxia conditions.

Abstract: A system, method and computer program product for solving a problem based on ontology methods for data/knowledge presentation and processing, may be implemented in a linguistic processing module. The basic components of the linguistic processing module may be a linguistic knowledge base (KB), an ontology KB, and/or an expert KB. The problem solved may include a user or an inventor problem. The method may include storing a user query which may include a structured, or a non-structured description, parsing a non-structured query to create a structured query including a formal semantic representation of the query, semantically expanding the formal semantic representation, searching the expert KB using the formal semantic representations to local solutions for the problem, and semantically sorting the solutions to produce a list of solutions of the problem.

Abstract: A method and systems that allows building cause effect tree related to initial problem statement. Each event of cause effect tree can be added with additional axes: hierarchy and operation. Wizards and templates provide simplify adding of undesirable effects. System automatically generates alternative problem statements and queries for search in knowledge bases.