8 results about "Local sequence" patented technology

The application provides an efficient deep learning method and system based on a squeeze-excitation-network and a ConNet network, which comprises the following steps: obtaining a protein global sequence and a protein local sequence as a sample set, and setting model parameters of a DeepNet deep framework; dividing the sample set into a training set and a verification set, setting a model architecture of the DeepNet deep framework, extracting effective feature information, combining negative samples and positive samples in the protein global sequence and the protein local sequence, and sending the samples into the DeepNet deep framework for training and hyperparameter tuning; adaptively coding the protein global sequence and the protein local sequence; designing a local sequence processing branch and a global sequence processing branch, extracting network structure features and key information between long and short sequences by using different scale convolution networks, and calculating final prediction probability according to the key information. The application solves the technical problems that sufficient feature information cannot be extracted and global information is not fully considered and between global information and local information.

The application discloses a high-precision time of arrival estimation method, equipment and medium for a burst frequency hopping communication system, relates to the field of time synchronization of a burst communication system, and comprises the following steps: a receiving end performs multi-channel down-conversion and 2 times symbol sampling on a frequency hopping signal, and performs correlation processing on a local sequence; non-coherent accumulation is performed on the correlation results after the correlation results are aligned according to a frequency hopping pattern and a delay difference, and a frame header stamping time of a peak time is recorded; a timing error estimation value is calculated by performing linear fitting on the correlation peak results; a corresponding compensation time is obtained by querying a time compensation table established offline, so as to correct the timing error estimation value, and a precise timing error estimation value is obtained; and finally, the precise estimation value is compensated to the frame header stamping time, and a high-precision time of arrival is output. The application realizes high-precision time synchronization under a low sampling rate, and effectively reduces hardware cost.

The present application relates to the technical field of communication, and especially relates to a correlation processing system, method and device for PSS detection and a medium. The system comprises: a parity separation module, configured to separate received air interface data into odd road sequences and even road sequences and output; two-way matched filters, one of which is configured to multiply the odd road sequences as input data with odd road data in a PSS local sequence to obtain odd road multiplication results, and the other of which is configured to multiply the even road sequences as input data with even road data in the PSS local sequence to obtain even road multiplication results; and a post-processing module, configured to first perform modulo operation on the odd road multiplication results and the even road multiplication results respectively, and then perform addition operation to generate correlation results. The scheme of the present application can effectively resist the detection performance decline caused by frequency offset, without the need for multiple attempts in the possible frequency offset range, greatly shortening the time consumed by PSS detection.

This invention relates to a method for extracting entity nesting from equipment maintenance and support text based on multi-technology fusion, belonging to the field of big data. The method includes: acquiring equipment maintenance and support text and performing word segmentation to obtain multiple sequence fragments; setting scanning windows of different lengths and scanning and filtering the sequence fragments according to configured filtering rules to obtain candidate sequences; inputting the candidate sequences into a BERT pre-trained model to convert them into feature vector sequences; using a BiLSTM neural network to extract features from the feature vector sequences and obtain sequence context information; using a TextCNN neural network to process the feature vector sequences and obtain local sequence features; concatenating the features to obtain fused features; mapping the fused features to label scores through a fully connected layer; calculating the label probability of a sequence fragment being an entity based on the label scores to predict the entity label corresponding to the sequence fragment. This invention's method can improve the efficiency and quality of equipment maintenance and support work by integrating multiple methods and strategies.

The application belongs to the technical field of deep reinforcement learning, and discloses a multi-agent multi-task cooperative reinforcement learning method based on a space-time fusion architecture, which comprises the following steps: step 1, initializing a task sampling probability, forming an entity embedding vector sequence and a task embedding vector; step 2, inputting the entity embedding vector sequence and the task embedding vector into a noise-resistant feature extraction layer to obtain deep time sequence features; step 3, generating a dynamic weight matrix in real time to obtain local Q values output by an agent; step 4, a Transformer hybrid network receiving a global state vector of an environment, a task embedding vector and a local Q sequence to give a global action value; step 5, calculating a total loss; and step 6, based on the total loss, training an optimization unit to update all network parameters, and simultaneously, judging whether a preset evaluation round is reached. The application has stronger anti-interference ability and more stable control performance, and realizes adaptive control of multiple heterogeneous tasks.

This application provides a graphical user interface (GUI) operation method, apparatus, device, and medium. The method includes: performing intent recognition and parsing on natural language commands to obtain a structured intent sequence; performing parsing and semantic mapping operations on the candidate GUI operation sequence to obtain a structured GUI operation sequence; constructing a matching score matrix between the structured GUI operation sequence and the structured intent sequence based on a local sequence alignment algorithm; determining the element with the largest value in the matching score matrix as the matching degree; and executing the candidate GUI operation sequence if the matching degree is greater than a threshold. This solves the semantic difference between the GUI operation sequence and the natural language commands, and robustly handles the mismatch in granularity and length between the two sequences, thereby improving the accuracy of the calculated matching degree and thus increasing the consistency between the GUI operation result and the user's expectations, enhancing the user experience.

The application belongs to the field of bioinformatics and computational biology, and particularly relates to a mutation detection method and system based on an attention mechanism and dynamic routing. The method comprises: obtaining a sequencing sequence alignment result, and extracting candidate mutation sites with high confidence through threshold setting preliminary screening; a local sequence window is intercepted with the candidate site as the center, a multi-dimensional feature tensor is constructed by fusing physical space position, positive and negative chain alignment classification and sequencing quality attributes; the multi-dimensional feature tensor is input into a backbone network, background noise is suppressed by using a Gaussian guided space attention mechanism, and long-distance multi-scale mutation features are extracted by using a shift window mechanism of a Swin Transformer and a dynamic routing strategy of a mixed expert module; finally, the existence of mutations, zygote states and allele sequences are determined in sequence by using a multi-task dynamic decoding network, and network parameter optimization and result output are completed by using a cost-sensitive focal loss. The application avoids the statistical feature loss problem caused by traditional image mapping, alleviates the limitation of limited local receptive field, realizes adaptive adjustment of the calculation path according to the sequence complexity by introducing a dynamic routing mechanism, overcomes the configuration imbalance of static models in the allocation of computing power and the capacity of local parameters, and enhances the detection accuracy and stability of the model under complex variation and low sequencing depth conditions.

The invention discloses a ribosome stagnation relative strength prediction method of a MambaPlus and Transform parallel architecture, and belongs to the crossing field of bioinformatics and artificial intelligence. According to the model, firstly, a DNA sequence is converted into multi-dimensional feature representation through K-mer and One-hot composite coding, a local sequence pattern is extracted by adopting a parallel multi-scale convolutional neural network, and long-range dependency and global context information are modeled respectively in combination with a MambaPlus encoder and a Transform encoder; multi-modal features are integrated through an adaptive weighted fusion strategy, dynamic aggregation is carried out on sequences in combination with attention pooling, and finally prediction of ribosome stagnation relative strength is realized in combination with biophysical features. The error of the depth model for predicting the relative strength of ribosome stagnation is smaller than 0.27, and compared with a traditional model, the depth model has higher accuracy.