Abstract: This disclosure relates to a method and an apparatus for training a text classification model. The method may include determining a semantic representation of the training sample using the text classification model and determining a predicted classification result of the training sample based on the semantic representation. The method may further include generating an adversarial sample corresponding to the training sample based on the training sample and perturbation information and determining a semantic representation of the adversarial sample corresponding to the training sample using the text classification model.

Abstract: The present application presents a method for adjusting antenna directions, an apparatus for adjusting antenna directions, a base station, an electronic device, and a computer-readable storage medium, and relates to the field of wireless communication technology. The method comprises: processing tidal time information of a tidal cell to obtain a plurality of tidal time periods, the tidal cell is a communication cell in which there is a tidal effect; determining a plurality of antenna weights corresponding to the tidal time periods based on the plurality of tidal time periods and neighbor cell information of the tidal cell; and adjusting directions of RF antennas of the tidal cell based on the tidal time periods and their corresponding antenna weights.

Abstract: The disclosure provides a liquid crystal display panel and a display apparatus. The liquid crystal display panel of the disclosure includes: first and second substrates assembled to form a cell, a plurality of main spacers therebetween, and an auxiliary spacer around at least a portion of the main spacers. Height of the auxiliary spacer is greater than or equal to that of the main spacer. The display panel further includes: pillows on side of the first substrate close to the second substrate and each abutting against a corresponding main spacer. An orthographic projection of the main spacer on the first substrate falls within an orthographic projection of the pillow on the first substrate, and an orthographic projection of the auxiliary spacer on the first substrate does not overlap with the orthographic projection of the pillow on the first substrate.

Abstract: A photo detection substrate, including: a base substrate; and a plurality of detection pixel units on the base substrate, where each detection pixel unit includes a signal reading circuit and a photoelectric conversion structure; the signal reading circuit includes at least one transistor each including a gate and an active layer pattern, the active layer pattern includes a channel region and a source/drain doped region; at least one transistor in the signal reading circuit is a superposed transistor on a side of the photoelectric conversion structure away from the base substrate, an orthographic projection of the active layer pattern of the superposed transistor on the base substrate overlaps an orthographic projection of the photoelectric conversion structure in the same detection pixel unit on the base substrate, and a material of the gate of the superposed transistor includes a transparent conductive material.

Abstract: The present disclosure provides a display substrate, a method for manufacturing the display substrate and a display panel. The display substrate includes: a first semiconductor layer on a base substrate, where an active layer of the first thin film transistor is in the first semiconductor layer, and the active layer of the first thin film transistor at least comprises a channel region and a drain contact region; an interlayer insulation layer on a side of the first semiconductor layer away from the base substrate; and a first conductive layer on a side of the interlayer insulation layer away from the first semiconductor layer, wherein the pixel electrode is located in the first conductive layer, and the pixel electrode in the pixel unit is directly and electrically connected to the drain contact region of the active layer of the first thin film transistor through a through hole.

Abstract: A thin film transistor and an ultrasonic imaging base board. The thin film transistor includes: a substrate (100), and a first gate (101), a first gate insulation layer (102), a first active layer (103), a second gate insulation layer (104) and a second gate (105) stacked on a side of the substrate (100), and the first gate (101) is connected to the second gate (105).

Abstract: The disclosure relates to an array substrate, a display panel, and a method for manufacturing an array substrate. The array substrate includes a first active layer on a substrate; a second active layer on a side of the first active layer away from the substrate; an intermediate layer between the first and second active layers and including a first via arriving at the first active layer, wherein the second active layer includes first, second and third sub-active layers, the first sub-active layer extending around a perimeter of the first via and in a direction away from the first via, the second sub-active layer covering a side wall of the first via, the third sub-active layer being at a bottom of the first via and covering a portion of a surface of the second active layer exposed by the first via, and wherein the second active layer is continuous.

Abstract: A computer device acquires, based on a prior network and a posterior network of a dialogue model, at least two first dialogue features and at least two second dialogue features of a first dialogue. The computer device outputs, using the prior network, a probability distribution of dialogue features in accordance with the acquired dialogue features and estimates, using the posterior network, the probability distribution of the dialogue features output by the prior network. The computer device updates the dialogue model and updates the posterior network based on dialogue features of the first dialogue. The computer device also updates a discriminator in the dialogue model based on at least two first dialogue features and at least two second dialogue features of a second dialogue. In accordance with a determination that the dialogue model has satisfied a training end condition, the computer device determines that the dialogue model is a trained model.

Abstract: The present disclosure provides a display substrate, including: a base substrate; and a plurality of spacers on a side of the base substrate and each comprising at least one first sub-pattern and at least one second sub-pattern stacked along a normal direction of the base substrate, the first sub-pattern being made of an organic material and the second sub-pattern being made of an inorganic material. The present disclosure also provides a method for manufacturing the display substrate and a display device.

Abstract: A method and apparatus for generating a hidden state in a recurrent neural network for language processing are provided. The method includes: generating regional word vectors of at least two dimensions of a target word vector inputted at a first moment; combining the regional word vectors to obtain combined regional word vectors of at least two dimensions; performing aggregation transformation processing on the combined regional word vectors based on a feedforward neural network, to obtain an aggregated word vector corresponding to the target word vector; and generating a target hidden state corresponding to the target word vector based on the aggregated word vector.

Abstract: Embodiments of this disclosure disclose a sentence processing method and device. The method may include performing word segmentation operation on a source sentence to be encoded to obtain m words. The method may further include obtaining an ith word in the m words using an ith encoding processing node in the n encoding processing nodes, and obtaining an (i?1)th word vector from an (i?1)th encoding processing node. The method may further include performing linear operation and non-linear operation on the ith word and the (i?1)th word vector using the first unit of the ith encoding processing node to obtain an ith operation result, and outputting the ith operation result to the at least one second unit for processing to obtain an ith word vector. The method may further include generating, in response to obtaining m word vectors, a sentence vector according to the m word vectors.

Abstract: The present invention provides a pixel unit structure, in which a source electrode is connected to a data line in a thin film transistor; the gate electrode is connected to the gate line; the drain electrode is disposed on a side of the gate insulating layer away from the substrate; the metal oxide semiconductor layer is disposed on a side of the gate insulating layer away from the substrate, and includes a semiconductor portion and a first conductive portion and a second conductive portion respectively located on both sides of the semiconductor portion; a terminal of the first conductive portion adjacent to the drain electrode is connected to the drain electrode or serves as at least a portion of the drain electrode; and the second conductive portion is connected to the source electrode through a first via formed correspondingly on the gate insulating layer and the interlayer dielectric layer.

Abstract: The disclosure provides a liquid crystal display panel and a display apparatus. The liquid crystal display panel of the disclosure includes: first and second substrates assembled to form a cell, a plurality of main spacers therebetween, and an auxiliary spacer around at least a portion of the main spacers. Height of the auxiliary spacer is greater than or equal to that of the main spacer. The display panel further includes: pillows on side of the first substrate close to the second substrate and each abutting against a corresponding main spacer. An orthographic projection of the main spacer on the first substrate falls within an orthographic projection of the pillow on the first substrate, and an orthographic projection of the auxiliary spacer on the first substrate does not overlap with the orthographic projection of the pillow on the first substrate.

Abstract: A display substrate includes: a base substrate including a display area and a frame area on at least one side of the display area; a first transistor in the frame area including a first active layer; a second transistor in the display area including a second active layer at a side of the first active layer facing away from the base substrate, the second active layer including a first channel region; and a light shielding layer in the display area. The light shielding layer contains the same element as the first active layer. A material of the light shielding layer can absorb at least part of ultraviolet light and transmit part of visible light. An orthogonal projection of the light shielding layer on the base substrate covers an orthogonal projection of the first channel region on the base substrate.

Abstract: A data processing method includes: obtaining a to-be-detected text, and determining a context word set and a candidate substitute word set corresponding to a to-be-detected word in the to-be-detected text to be inputted into a pronoun resolution neural network for feature extraction; performing positive-example iteration processing and negative-example iteration processing on the features corresponding to the context word set and the candidate substitute word set, to obtain a positive-example feature vector length and a negative-example feature vector length, and calculating a substitute probability corresponding to each candidate substitute word in the candidate substitute word set according to the positive-example feature vector length and the negative-example feature vector length; determining a target substitute word according to the substitute probability corresponding to the each candidate substitute word; and inserting the target substitute word into the to-be-detected text according to a position corr

Abstract: The present invention provides an antibody binding a human tumor-associated calcium signal sensor 2 (Trop-2) protein or fragments thereof, and use of the antibody or fragments thereof in preventing or treating diseases. The antibody or fragments thereof of the present invention can effectively bind to the human Trop-2 protein, and have internalization activity, and the internalization activity is enhanced after ADC drug labeling, and the in vivo efficacy and safety of a mouse model are not lower than those of a control antibody.

Abstract: In view of the technical problem that TGF-? RH and a fusion protein thereof are prone to degradation and breakage during the process of recombinant expression, provided are a TGF-? RII mutant and a fusion protein thereof. Compared with the extracellular region of the wild-type TGF-? RII of SEQ ID NO: 6, the TGF-? RII mutant has a mutation selected from Gln at position 6, Asp at position 12 and Gly at position 20. The TGF-? RII mutant is capable of binding to TGF-?. Compared to wild-type TGF-? RII, the TGF-? RII mutant has less breakage and/or degradation during recombination expression. It is convenient for large-scale production of antibodies/the TGF-? RII bifunctional protein with a more stable quality.

Abstract: A language sequence labeling method and includes: reading a first embedding representation of a language sequence, the first embedding representation including a character-level word embedding representation, a pre-trained word embedding representation, and a global word embedding representation of the language sequence, the global word embedding representation referring to a global context representation of the language sequence; performing first depth transformation (DT) encoding on the first embedding representation based on a first DT recurrent neural network (RNN), to output a first hidden-layer state representation corresponding to each word in the language sequence; and decoding the first hidden-layer state representations of the language sequence, to obtain a labeling result of one or more elements in the language sequence.

Abstract: The present invention provides an antibody binding to programmed cell death protein 1 (PD-1) or a fragment thereof and a use of the antibody or a fragment thereof in preventing or treating a tumor or cancer. The antibody or a fragment thereof of the present invention can effectively block the interactions of PD-1/PD-L1 and PD-1/PD-L2 and has a unique antigen binding epitope and unique properties in terms of efficacy and safety.

Abstract: This disclosure relates to a method and an apparatus for training a text classification model. The method may include determining a semantic representation of the training sample using the text classification model and determining a predicted classification result of the training sample based on the semantic representation. The method may further include generating an adversarial sample corresponding to the training sample based on the training sample and perturbation information and determining a semantic representation of the adversarial sample corresponding to the training sample using the text classification model.