Abstract: A multi-lead electrocardiogram (ECG) signal classification method based on self-supervised learning relates to the technical field of ECG signal classification. The method includes: processing an original signal through different data augmentation methods, designing an appropriate encoder module, extracting a feature of an ECG signal through a large amount of easily available unlabeled data such that an encoder learns more class information of the ECG signal, fine-tuning the model encoder with a small amount of labeled data for feature optimization, and continuously optimizing a parameter of a feature extractor by training a model such that a generated feature well reflects a structure and information of input data. Through self-supervised learning, the method reduces obstacles caused by performing ECG signal classification through a large amount of expensive manually labeled data, improving the generalization ability of the model.

Abstract: A multi-lead electrocardiogram (ECG) signal classification method based on self-supervised learning relates to the technical field of ECG signal classification. The method includes: processing an original signal through different data augmentation methods, designing an appropriate encoder module, extracting a feature of an ECG signal through a large amount of easily available unlabeled data such that an encoder learns more class information of the ECG signal, fine-tuning the model encoder with a small amount of labeled data for feature optimization, and continuously optimizing a parameter of a feature extractor by training a model such that a generated feature well reflects a structure and information of input data. Through self-supervised learning, the method reduces obstacles caused by performing ECG signal classification through a large amount of expensive manually labeled data, improving the generalization ability of the model.

Abstract: An electrocardiograph (ECG) signal detection and positioning method based on weakly supervised learning is provided. A deep learning model mainly includes a multi-scale feature extraction module, a self-attention encoding module, and a classification and positioning module. An extracted original ECG signal is denoised and segmented to obtain a fixed-length pure ECG signal segment. In the convolutionally-connected multi-scale feature extraction module, a channel local attention (CLA) layer is introduced, and a PReLU activation function is used to achieve a better local information extraction capability. The self-attention encoding module is introduced to establish an association between a local feature and a global feature. The classification and positioning module is introduced to output a general location of an abnormal signal. A fusion module enables the model to map a local predicted value onto a global predicted value, and model parameters are trained on a weakly annotated dataset.

Abstract: The present invention belongs to the field of new material technology and relates to a bismuth/bismuth titanate heterojunction hollow nanospheres and methods for preparing the same and applications thereof. The bismuth/bismuth titanate heterojunction hollow nanospheres is obtained by adding a bismuth salt and tetrabutyl titanate into an organic solvent, mixing evenly, adding an inorganic alkali solution to generate a precipitation, then sequentially adding an organic amine and an organic acid, performing a solvothermal treatment to obtain bismuth titanate hollow nanospheres, and then thermally reducing the bismuth titanate hollow nanospheres with a borohydride under an inert atmosphere. The bismuth/bismuth titanate heterojunction hollow nanospheres provided not only exhibit significant photocatalytic activity in degrading tetracycline hydrochloride but also enable a tighter binding of the metal to the photocatalytic material, thus offering superior structural stability.

Abstract: A bismuth ferrite film material, a method for integrally preparing a bismuth ferrite film on a silicon substrate at a low temperature, and an application, includes: magnetron sputtering a bottom electrode, a buffer layer and a bismuth ferrite film on one surface of a Si substrate in sequence from bottom to top at a processing temperature of 300-400° C.; reducing the temperature to room temperature; and a top electrode is deposited via magnetron sputtering on the surface of the bismuth ferrite film; the buffer layer mentioned hereof is a conductive oxide which matches the lattice of bismuth ferrite and is of a perovskite structure (ABO3). According to the present invention, the temperature for preparing the bismuth ferrite film material can be reduced to 450° C. or below, and the bismuth ferrite film material has a high spontaneous electric polarization.

Abstract: A multi-label electrocardiogram (ECG) signal classification method based on an improved attention mechanism is provided. A model is constructed for classifying a multi-label (multi-lead) ECG signal. The model has a strong ECG data learning ability, allowing a computer to fully extract a feature of the ECG signal and construct a data processing channel model. Therefore, the multi-label (multi-lead) ECG signal can be effectively classified, improving the accuracy and precision of classification.

Abstract: A peony stamen-dodder composite, which is prepared from 10-18 parts by weight of peony stamen and 5-12 parts by weight of dodder. In the preparation process, the peony stamen is sequentially subjected to desensitization, impregnation with a saline and drying, and the dodder is impregnated with a saline and baked. The pre-treated peony stamen and dodder are crushed, impregnated in warm water, ultrasonicated and filtered, and the filter residue is added with water, ultrasonicated and filtered. The filtrates are combined and concentrated under vacuum to obtain the desired composite. This application further provides an application of the composite in the treatment of nephritis.

Abstract: A material with pine-branch like samarium oxide/graphene/sulfur gel structure, and a preparation method and use thereof. The material is reduced graphene oxide carrying pine-branch like samarium oxide on the surface to form a cross-linked gel structure, and sulfur is loaded on the gel structure. The preparation method includes: subjecting graphene oxide and a samarium salt to hydrothermal reduction to prepare a reduced graphene oxide/samarium precursor; under an inert atmosphere, thermolysing the reduced graphene oxide/samarium precursor to obtain a reduced graphene oxide/pine-branch like samarium oxide gel; and melting and diffusing the sulfur onto the reduced graphene oxide/pine-branch like samarium oxide gel. The material with pine-branch like samarium oxide/graphene/sulfur gel structure greatly improves the electrochemical performance of lithium-sulfur batteries.

Abstract: The present invention relates to a bismuth tungstate/bismuth sulfide/molybdenum disulfide heterojunction ternary composite material and a preparation method and application thereof. The composite material is composed of bismuth tungstate, bismuth sulfide and molybdenum disulfide in an ordered layered way, Bi2WO6 is an orthorhombic system, Bi2S3 is a p-type semiconductor located on a (130) crystal face, MoS2 is a layered transition metal sulfide located on a (002) crystal face, the whole composite material is of a spherical structure with an unsmooth surface, and a layer of nanosheets uniformly grow on an outer layer. The average particle size of composite materials is in the range of 2.4-2.6 ?m. The spherical Bi2WO6/Bi2S3/MoS2 heterojunction ternary composite material prepared in the present invention has good adsorption of Cr(VI) and high catalytic reduction ability under visible light.

Abstract: An electrocardiograph (ECG) signal quality evaluation method based on a multi-scale convolutional and densely connected network is provided. Firstly, an original ECG signal is preprocessed to remove a baseline drift and power line interference. Then, based on a consistency principle of a label determining result and a principle of setting a confidence coefficient, an AlexNet model is trained to mutually correct incorrect labels in a dataset to obtain a final ECG signal fragment for quality classification. Finally, the signal fragment is input into an improved lightweight densely connected quality classification model to classify quality of the ECG signal fragment.

Abstract: A method for preparing pinacolone is provided. Raw materials including 2-methyl-2-butene, hydrochloric acid and formaldehyde are reacted in the presence of a catalyst to produce pinacolone. The catalyst is a single lanthanide Lewis acid, a compounded lanthanide Lewis acid or a lanthanide metal oxide soluble in hydrochloric acid. The lanthanide Lewis acid is lanthanum chloride, cerium chloride, praseodymium chloride, neodymium chloride, erbium chloride, holmium chloride, dysprosium chloride or thulium chloride. The method is performed through a continuous two-step reaction. In the first reaction, 2-methyl-2-butene and hydrochloric acid are reacted to form an intermediate, which undergoes a second reaction with formaldehyde in the presence of the catalyst to produce pinacolone.

Abstract: A method for separating biomass components by a ternary system, which relates to a technical field of biomass separation, and includes the following steps of: cooking and separating a biomass raw material by using a cooking liquor consisting of organic acid, small aromatic nucleophilic organic molecule and hydrogen peroxide to obtain solid residue and extracting solution, washing and screening the solid residue to obtain paper pulp, and separating and extracting lignin and/or hemicellulose from the extracting solution. This cooking system could effectively minimize the content of residual lignin and other compounds with chromophore groups in the obtained pulp, directly producing the high-whiteness pulp with excellent performance without additional bleaching process. In addition, the hemicellulose saccharides and high-activity lignin can be also obtained, so that the method has an important significance for realizing high value and industrialization of biomass resource utilization.

Abstract: A lightweight attribute-based signcryption (ABSC) method for cloud-fog-assisted Internet-of-things: performing, by a central authority, system initialization to generate a system key pair, and disclosing a public key, the public key including a symmetric encryption algorithm (SEA) and a key derivation function (KDF); generating, by the central authority, a decryption key and an outsourcing decryption key based on a decryption attribute set of a data user, and generating a signature key and an outsourcing signature key based on a signature access structure; calling, by a data owner, a fog node for outsourcing signature, performing symmetric encryption on a plaintext based on a symmetric key, and performing ABSC on the symmetric key based on a defined encryption access structure; and calling, by the data user, a fog node for outsourcing signature verification, calling a fog node for outsourcing decryption, and performing symmetric decryption on a ciphertext based on an outsourcing decryption result.

Abstract: An ABE method with multiple tracing attribute authorities: performing, by a central authority, system initialization to generate a public parameter and disclosing the public parameter; performing, by each of attribute authorities, initialization to generate a key pair, and disclosing a public key in the key pair; performing, by a data owner, symmetric encryption on plaintext data, performing ABE on a symmetric key based on a hidden access structure, and generating an integrity verification value; requesting, by a data user, a decryption key to the attribute authority according to an own attribute; restoring, by the data user in response to decryption, an access structure, generating an outsourcing decryption key, sending the outsourcing decryption key to a cloud storage center for semi-decryption; generating, by the cloud storage center, a semi-decrypted ciphertext, and feeding the semi-decrypted ciphertext back to the data user; fully decrypting the semi-decrypted ciphertext according to a private decryption

Abstract: An electrocardiograph (ECG) signal quality evaluation method based on a multi-scale convolutional and densely connected network is provided. Firstly, an original ECG signal is preprocessed to remove a baseline drift and power line interference. Then, based on a consistency principle of a label determining result and a principle of setting a confidence coefficient, an AlexNet model is trained to mutually correct incorrect labels in a dataset to obtain a final ECG signal fragment for quality classification. Finally, the signal fragment is input into an improved lightweight densely connected quality classification model to classify quality of the ECG signal fragment.

Abstract: A facial expression-based detection method for deepfake by generative artificial intelligence (AI) constructs an AIR-Face facial dataset for generative AI-created face detection training, and uses an untrained information feature space for real and fake classification. Nearest linear detection is performed in this space to significantly improve the generalization ability of detecting fake images, especially those created by new methods such as diffusion models or autoregressive models. The detection method improves the performance of extracting features of generative AI-created faces through phased trainings, and detects generative AI-created faces through the feature space. Compared with other methods, the detection method scientifically and effectively improves the accuracy of generative AI-created face recognition, and fully mines the potential semantic information of generative AI-created faces through phased trainings.

Abstract: The present invention provides a self-healing ceramic material with reduced porosity and a method for preparing the same. The self-healing ceramic material comprises the following components by volume: 60-85 parts by volume of Al2O3, 10-20 parts by volume of TiN, 10-20 parts by volume of TiSi2, 0.1-1 parts by volume of MgO, and 0.1-1 parts by volume of Y2O3. Wherein, Al2O3 serves as a matrix, TiN and TiSi2 act as repair agents, and MgO and Y2O3 function as sintering aids. The repair function is realized by adding TiN and TiSi2, which have repair capabilities, to the ceramic matrix, enabling the ceramic material to heal cracks. TiN plays the main role of repair function, and TiSi2 assists in the repair and helps reduce the formation of surface pores.

Abstract: An electrocardiograph (ECG) signal detection and positioning method based on weakly supervised learning is provided. A deep learning model mainly includes a multi-scale feature extraction module, a self-attention encoding module, and a classification and positioning module. An extracted original ECG signal is denoised and segmented to obtain a fixed-length pure ECG signal segment. In the convolutionally-connected multi-scale feature extraction module, a channel local attention (CLA) layer is introduced, and a PReLU activation function is used to achieve a better local information extraction capability. The self-attention encoding module is introduced to establish an association between a local feature and a global feature. The classification and positioning module is introduced to output a general location of an abnormal signal. A fusion module enables the model to map a local predicted value onto a global predicted value, and model parameters are trained on a weakly annotated dataset.

Abstract: An electrocardiograph (ECG) signal enhancement method based on a novel generative adversarial network (GAN) effectively enhances a capability of a generator model for understanding and expressing input data by using a multi-branch structure of bi-directional long short-term memory (BiLSTM) neural networks with different quantities of hidden neurons, and stitching outputs of last time steps of forward propagation of the different BiLSTM networks. A new ECG signal enhancement module EEA-Net is proposed, which uses an adaptive convolutional layer to dynamically adjust a size of a convolution kernel, making the model more flexible in processing input sequences of different lengths. In addition, the model uses an adaptive average pooling layer to perform weighted average pooling on the input data to better capture important information of the input data.

Abstract: A multi-label electrocardiogram (ECG) signal classification method based on an improved attention mechanism is provided. A model is constructed for classifying a multi-label (multi-lead) ECG signal. The model has a strong ECG data learning ability, allowing a computer to fully extract a feature of the ECG signal and construct a data processing channel model. Therefore, the multi-label (multi-lead) ECG signal can be effectively classified, improving the accuracy and precision of classification.