5results about How to "Easy to decode" patented technology

An experimental platform for implantable brain-computer interface of limited movement, comprising: a cooperative control module, which extracts the brain neuron signals of a limited rat through a microelectrode, pre-processes and extracts features of the neuron signals, analyzes and processes the movement information contained therein, converts the movement information into appropriate control instructions, and delivers the control instructions from Raspberry Pi to Arduino to control the corresponding actions of a machine rat made by the limited rat, so as to realize the movement control and sensing feedback between the limited rat and the machine rat; and an external control module, which is installed on the machine rat, and is used for carrying out epidural spinal cord electrical stimulation on the limited rat with partial movement function after incomplete spinal cord injury and combining with external equipment for rehabilitation training. The application can analyze and record the neuron signals of specific regions when the limited rat performs actions, which is helpful for in-depth analysis of the neuron activity characteristics and regulation mechanism of the limited rat.

The present application relates to the technical field of communication, especially to a processing method of QC-LDPC decoder, the method comprising: obtaining soft information of QC-LDPC decoder; performing field splicing on the soft information to obtain spliced code block set; performing decoding on the spliced code block set to obtain decoding information field; deleting zero-puncturing bits in the decoding information field to obtain standard information field. The method improves the working efficiency of QC-LDPC decoder, makes QC-LDPC decoder compatible with different code length and code rate of QC-LDPC code, and further makes QC-LDPC decoder work flexibly and in real time according to different protocols.

The invention discloses a framing transmission method based on byte stream communication. The method comprises the steps that a sending buffer transmits data attributes of byte stream data to a coding resource allocator; the coding resource allocator calculates a storage space required after coding based on the data attribute of the byte stream data; the frame stream encoder encodes the byte stream data based on the calculated storage space required after encoding to obtain encoded data, and then the encoded data is transmitted to target end equipment through the byte stream transmitter; the byte stream receiver receives coded data of target end equipment, and the decoding resource allocator calculates a storage space required after decoding based on data attributes of the coded data; and the frame stream decoder decodes the coded data based on the calculated storage space required after decoding to obtain reduced data, and stores the reduced data in a receiving buffer. According to the invention, rapid framing transmission is realized through a one-byte seven-bit transmission mechanism, so that stable and reliable framing of data is maintained under high-frame-rate transmission, and the transmission efficiency is improved.

The application discloses an event segmentation and extraction method and system in a text extraction task, comprising the following steps: adopting a data segmentation and labeling strategy of a Span fragment, building a neural network segmentation algorithm model based on a pre-training model, enabling the network to learn the first and last token positions of an event, segmenting the event, and reducing the number of events processed by the extraction model at a time; building a neural network extraction algorithm model based on a pre-training model, converting the event extraction task into three subtasks of argument recognition, event division and graph search. The above method can avoid overly complex decoding of the extraction model, improve the accuracy of text event extraction, and reduce extraction errors. In addition, the application is simple and easy to implement, adopts a segmentation and extraction Pipeline mode, greatly improves the accuracy of text event extraction, has high extraction efficiency, and theoretically does not have the exposure bias problem of a traditional autoregressive model.

The present disclosure relates to a method, system, electronic device and storage medium for handwritten formula recognition, the method comprising the following steps: encoding a handwritten formula image into a feature map, performing two-dimensional position encoding on a feature vector in the feature map to obtain a position code, adding the position code to the feature vector to obtain a feature image code; using a semi-autoregressive tree decoder to decode a character layout tree corresponding to the handwritten formula, obtaining a relationship between character nodes of the character layout tree, and introducing the relationship between character nodes into a decoder model; using two auxiliary training tasks to jointly participate in model training, combining the auxiliary training task loss function with the cross entropy loss function of the decoder model, obtaining a total loss function of the decoder model, and using an optimizer to train the decoder model to obtain a complete decoder model. The handwritten formula recognition method disclosed in the present disclosure enhances the generalization ability of the model and improves the accuracy of the handwritten mathematical formula recognition task.