93 results about "End point detection" patented technology

The invention relates to the technical field of voice recognition intelligent home control, and discloses a non-specific person voice recognition intelligent switch control method and system based on deep learning. The non-specific person voice recognition intelligent switch control method is applied to intelligent switch control equipment, and specifically comprises the following steps of S101, receiving original audio signals continuously collected in a to-be-controlled environment, and preprocessing the collected original audio signals, and then a starting point and an ending point of an effective voice segment are positioned by adopting endpoint detection based on a double-threshold method and combining the characteristic parameters of the short-time energy and the short-time zero-crossing rate. A multi-layer hidden layer structure with Dropout regularization is adopted in a neural network model, the generalization ability of the model is enhanced, a context sensing mechanism is introduced into a semantic understanding module, a composite instruction containing azimuth information can be intelligently analyzed, crossing from recognition to understanding is achieved, and the method has the advantages of being high in practicability and easy to popularize. The system is ensured to maintain a high recognition rate for voice instructions of different users under different environment conditions.

Methods and systems for process chamber qualification for maintenance process endpoint detection are provided. Data collected by one or more sensors of a process chamber of a manufacturing system is identified. The identified data is collected during performance of initial maintenance operation(s) of a maintenance process. A current state of the process chamber is determined, based on the identified data, after the performance of the initial maintenance operation(s) based on the identified data. In response to a determination that the current state does not satisfy one or more chamber maintenance criteria, a set of subsequent maintenance operations to be performed to cause the current state of the process chamber to satisfy the criteria is identified. Performance of the set of subsequent maintenance operations is initiated at the process chamber.

Information processing device that includes: an information acquisition unit that acquires reliability degradation factor state information in a chemical mechanical polishing process of a substrate performed by a substrate processing device, the reliability degradation factor state information including at least one of wear state information indicating a wear state of components of the substrate processing device and processing state information indicating a processing state during polishing; and a state prediction unit that predicts reliability information of a polishing endpoint detection function for the reliability degradation factor state information by inputting the reliability degradation factor state information acquired by the information acquisition unit into a learning model that has been trained through machine learning to learn a correlation between the reliability degradation factor state information and reliability information of the polishing endpoint detection function that indicates reliability of an endpoint detection function that detects that the chemical mechanical polishing process has reached an endpoint.

The invention relates to the technical field of medical information processing, in particular to a medical voice instruction semantic analysis and form automatic filling system, which comprises the following modules: a voice recognition and analysis module, which is used for collecting a medical voice instruction, carrying out noise reduction processing and endpoint detection on a voice signal, and sending the voice signal to an automatic form filling module; generating a recognition text with confidence by using a voice recognition model customized by a medical scene, and performing named entity recognition and slot extraction on the recognition text to generate candidate semantic units; and the constraint generation module is used for extracting clinical logic constraints of the form fields according to the candidate semantic units. According to the method, constraint semantic representation is generated in combination with violation degree calculation, matching is completed under logic constraints by using a global optimization algorithm, and high-precision mapping of medical voice instructions and form fields is realized; not only is the integrity of semantic information ensured in the recognition stage, but also mistaken filling is effectively avoided through clinical logic constraints in the matching stage.

The application relates to a container endpoint detection method and device, a storage medium and equipment. A least square method is used to construct a straight line equation of n sampling points of laser scanning data of a to-be-detected container. Whether a sampling point after the nth sampling point belongs to the same straight line equation is judged to obtain an offset sampling point. The coordinate information of the first sampling point of the straight line equation and the coordinate information of the offset sampling point are used to obtain the coordinate information of a straight line endpoint. The straight line equation of the sampling point after the offset sampling point is reconstructed, and the coordinate information of the endpoint of the straight line segment corresponding to the straight line equation is obtained. Therefore, the position information of each endpoint of the to-be-detected container can be obtained according to the coordinate information of the endpoint, the user can quickly locate the endpoint position of the container according to the position information of each endpoint, and the detection efficiency of the container is improved.

The semantic perception real-time voice endpoint detection method is applied to a voice interaction system, and comprises the following steps: receiving a voice stream of a user and a voice stream sent by the system from at least two audio channels; performing causal voice coding on the voice stream of the user and the voice stream sent by the system to obtain corresponding double-path voice representation; a causal two-way modeling architecture is constructed, and the causal two-way modeling architecture is realized by adopting an autoregressive prediction network and is used for receiving two-way voice representation and modeling a speech round conversion dynamic state between a user and a system through an embedded causal constraint mechanism and a cross-channel interaction mechanism; based on the output of the autoregressive prediction network, jointly predicting the category of a speech round conversion event and the remaining time from the current moment to the end of the speech round; and a smoothing processing mode is used to correct the predicted speech wheel conversion event type, and a final endpoint detection result is determined. According to the method, the endpoint detection stability in a complex dialogue scene can be improved.

This invention discloses a method for accurate audio warning identification based on hybrid features. This method analyzes collected audio speech to determine the issuance time of the warning signal, thereby accurately evaluating the timeliness of the audio warning. First, a double noise reduction method using logmmse-spectral subtraction is employed to filter out noise information in the recorded speech. Next, endpoint detection based on short-time energy is used to mark the effective speech segments in the test speech. Then, MFCC features and waveform polynomial features are extracted from each frame of the effective speech segment. Subsequently, the two features are used as inputs to two channels of a convolutional neural network, and the outputs of the two channels are summed to obtain the hybrid features. Finally, the hybrid features are used as input to a softmax function, and the speech segment containing the target speech (audio warning signal) is determined by the maximum probability value. The starting position of this speech segment is the issuance time of the warning signal.

To provide a method for manufacturing a polishing pad with an endpoint detection window (window member) that exhibits excellent adhesion between the back surface of the window member and the adhesive surface of the second sheet. [Solution] A polishing pad comprising a first sheet having a first through-hole, a second sheet having a second through-hole, and a window member, wherein the first sheet has a polishing surface for polishing an object to be polished, the second sheet is bonded to the surface of the first sheet opposite to the polishing surface, the outer periphery of the second through-hole is formed to cover the outer periphery of the first through-hole, and the window member is positioned inside the first through-hole and on the second sheet such that the uppermost part of the window member is lower than the polishing surface; and a method for manufacturing a polishing pad comprising the steps of: preparing the polishing pad, and placing a spacer of a predetermined thickness on a press platen under the polishing pad around the polishing pad, and pressing the polishing pad with a press machine.

The application provides a regulation voice segmentation method based on speech recognition and endpoint detection, which is applied to air traffic control voice audio stream segmentation, and comprises the following steps: step 1, constructing a punctuation model based on speech recognition and a speech endpoint detection model; step 2, using the punctuation model based on speech recognition to recognize the audio data stream of the regulation voice, and outputting the corresponding text and sentence end identifier of the audio data stream; step 3, using the speech endpoint detection model to judge the speech starting point and ending point contained in the audio data stream of the regulation voice; step 4, segmenting the audio data stream of the regulation voice into audio segments; and step 5, applying the audio segments as data materials to the speech recognition process of the air traffic control system. Through the combination of speech recognition and endpoint detection, the application realizes the automatic segmentation of the air traffic control voice audio stream, and improves the accuracy and efficiency of the segmentation.

A technique for threat response associated with an endpoint detection and response (EDR) system. The system uses a combination of automated observable detection, threat intelligence enrichment, graph

The invention discloses a livestock respiratory rate estimation method, and relates to the technical field of animal health monitoring. The method comprises the following steps: collecting a livestock lung sound stream, segmenting into segments according to a fixed window length, and carrying out initial noise reduction; framing endpoints of the segments are detected, starting points and ending points of inspiration segments are marked, and after a sequence is obtained, a small subset and a large subset are divided according to the number of the inspiration segments; screening the effective inspiration section by using the expiration average duration time in the less set, and screening the effective inspiration section by using the difference value between the time interval of two adjacent inspiration sections and the expiration average duration time and the standard difference value in the more set; after the effective inspiration sections are screened out, the breathing frequency is calculated according to the ratio of the time interval between the head and tail inspiration sections of the target section to the number of effective breathing cycles. The accuracy of livestock respiratory rate detection can be effectively improved.

The application provides a speech endpoint detection method, device and equipment and a storage medium. The speech endpoint detection method comprises the following steps: obtaining target data, wherein the target data comprises target audio data; inputting the target data into a pre-trained audio frame classification model to obtain a classification result of an audio frame of the target audio data, wherein the audio frame classification model is trained by taking first audio data with noise in one or more categories of a plurality of noise categories as a training sample, taking a real category of the audio frame of the first audio data in a plurality of dimensions as a sample label, and the plurality of dimensions comprise a speech dimension and noise dimensions corresponding to the plurality of noise categories; and determining a speech endpoint according to the classification result of the audio frame of the target audio data. The speech endpoint detection method provided by the application can detect an accurate speech endpoint.

A method of determining an End Point Detection (EPD) time and an electronic device for performing the method are provided. The electronic device includes a processor and a memory electrically connected to the processor and configured to store commands executed by the processor. The processor may be configured to recognize an utterance when the commands are executed, determine an End Point Detection (EPD) time for terminating recognition of the utterance based on utterance data stored in the memory, and terminate the recognition of the utterance when the utterance is not additionally input within the EPD time. The utterance data may include sentence data, which includes a pattern related to a sentence type of the utterance, and word data, which includes a word of the utterance.

The application relates to the technical field of semiconductor detection, and discloses an end-point detection device capable of preventing free jet flow and entrainment and a dynamic adjustment method, wherein the detection device comprises a light source module, a gas chamber module, a variable-diameter module, a segmented temperature control module and a detector module; the gas chamber module comprises coaxially nested first, second and third cylinder bodies, a first chamber is formed between the first and second cylinder bodies, the inner cavity of the third cylinder body is a second chamber, a third chamber is formed between the third and second cylinder bodies, the inner wall of the first cylinder body and the outer wall of the second cylinder body are provided with wall-attached flow guide structures, and the third cylinder body is densely provided with exhaust holes; the variable-diameter module is used for adjusting the flow area of an exhaust pipe; and the segmented temperature control module is used for forming a temperature gradient which decreases along the gas flow direction. Through the cooperation of the anti-entrainment gas chamber structure, the variable-diameter module and the segmented temperature control module, the jet flow entrainment effect can be inhibited, and the accuracy of gas concentration measurement and the reliability of clean process monitoring can be improved.

The embodiment of the present application provides a process step endpoint detection method and a semiconductor process equipment, the method comprises the following steps: in the process of executing a process step, obtaining an executed process time and an expected process time of the process step; according to the expected process time and the executed process time, calculating a delay end detection time of the process step; according to the delay end detection time, triggering a first timing monitor to start timing monitoring, and triggering a skip processing for the process step when the timing time of the first timing monitor reaches the delay end detection time. The flexible variation of the process step endpoint time can be realized according to the expected process time set by the user, and the actual execution time of the process step can meet the expected process time of the user; when the timing time of the timing monitor reaches the delay end detection time, the skip can be automatically executed, and the button of manual end detection does not need to be manually triggered, so that the automation is higher.

A computer-implemented method for generating synthetic data describing a sample may include: processing, by a system operatively coupled to a processor, sample information describing a set of original features of the sample, thereby producing processed model input information; and generating composite image data by analyzing a model and based on the processed model input information, the composite image data being usable to generate a composite image and describing at least a portion of the set of original features. The computer-implemented method may also include generating, by the system, a composite image of dummy samples based on the composite image data, the composite image having at least one dummy feature different from and based on the set of original features.

Systems and methods of persistent file protection integrate cloud-based backups with EDR file protection. Use of local persistent file cache (PFC) data with cloud-stored file segments, facilitates reconstruction of file states from multiple sources. File storage and retrieval efficiency are optimized, even when local session caches are invalidated or when extensive file modifications exceed storage quotas.

This application provides a wafer processing system, an endpoint detection device, a method, and an electronic device. The wafer processing system includes: a wafer polishing assembly, including a polishing disk; an optical assembly disposed within the polishing disk and rotating synchronously with the polishing disk, used to acquire the reflection spectral signal of the wafer film layer in real time during the wafer polishing process; and a controller, communicatively connected to the optical assembly, used to select a matching spectral data processing strategy based on the current wafer polishing parameters to identify whether the reflection spectral signal is an abnormal spectrum; if not, to determine the current film layer thickness based on the reflection spectral signal; and to perform polishing endpoint determination based on the film layer thickness. The solution of this application can avoid misjudgment of the polishing endpoint caused by abnormal spectral data, ensuring the consistency and yield of wafer processing.

Disclosed is a method and device for image-based malware detection, and artificial intelligence-based endpoint detection and response system using same. The method includes the steps of: reading, as bytes of an unsigned integer of constant bits, an executable file or at least a portion of the code of an executable file located in a file path of a generated file obtained from a file generation event log detected in an EDR system or a processor, a memory, or a storage device associated with an endpoint, and converting same into a byte array; reading two bytes at a time from the byte array; using the two bytes of data as coordinates of an image of a preset size to increase a corresponding coordinate value, and adding an RGB channel to store an image array; and performing deep learning analysis using an image deep learning model on the image array.

The disclosed configurations generate synthetic attack data that emulates the markers of a cyberattack. The generated synthetic attack data is used to train an attack detection machine learning model that detects and mitigates actual cyberattacks in real-time. Synthetic attack data is generated by a synthetic attack data generation model, which is trained with a synthetic attack data generation prompt. The synthetic attack data generation prompt is constructed out of attack data samples and a prompt guideline. The prompt guideline is created from attack procedure descriptions, such as security blog posts or other write-ups about actual cyberattacks. Prompt guidelines may include samples of actual attack data that indicate how to format synthetic attack data. Once deployed, the attack detection machine learning model infers the occurrence of a cyberattack from log entries. Detected cyberattacks may be mitigated in an automated or semi-automated manner.

The invention discloses an early language ability evaluation and improvement method and system based on deep learning, relates to the technical field of child language development evaluation, and proposes the following scheme: obtaining discontinuous voice stream data of a child, carrying out endpoint detection through a pre-trained deep neural network model, and segmenting to generate voice segments, and recognizing each voice segment and incorporating the voice segments into different types of grammar slots, and combining the voice segments in the same grammar slots into a competition node set. According to the method, the incidence matrix is constructed by taking the co-occurrence probability value between the competition node sets of the adjacent syntax slots as the element, the competition node combination with the co-occurrence probability value greater than the preset co-occurrence threshold value is screened, and the initial logic path is generated through the Cartesian product operation, so that the problem of path stiffness caused by a fixed rule is solved; the probability-driven dynamic path generation is realized, so that the language logic path is closer to the actual language use probability and habit of children, and the authenticity and individuation level of evaluation are improved.

The invention discloses a process camouflage attack detection method based on a kernel object causal atlas and a graph neural network, and relates to the technical field of network security and deep learning. The invention aims to solve the problem that an existing endpoint detection system seriously depends on process names and static characteristics and is difficult to identify advanced attacks such as sequence imitation and process identity disguise. The method mainly comprises the following steps: firstly, capturing an input / output request packet (IRP) of a file system in real time through a kernel-level filter driver (Minifilter), and extracting multi-dimensional features including a process identifier, an access permission mask, an I / O control code and a file object pointer (File Object Pointer); secondly, providing a causal atlas construction technology based on a file object pointer, aggregating discrete and cross-process I / O (Input / Output) operation into a logically coherent causal chain through a shared kernel object, and forcibly associating a disguised attack fragment with system background noise; and finally, constructing a graph convolutional neural network (GCN) model, performing feature aggregation and convolution calculation on graph nodes by using a time sequence edge and a causal edge, and identifying abnormal nodes of which behavior semantics are not consistent with claimed identities. According to the method, disguising of the process PID and the name can be effectively penetrated, the attack path is precisely restored through the bottom kernel logic, and the detection accuracy and robustness of ransomware, APT attacks and file-free attacks are remarkably improved.

A new approach is proposed to support integration of EDR data from a plurality of EDR tools / sources into an Open XDR framework in an automated manner. First, EDR data generated by each of the plurality of EDR tools covering a plurality of assets is ingested into the Open XDR framework. The ingested EDR data is then normalized through a unified EDR data model. The normalized EDR data is further enriched with one or more new data fields to better correlate the EDR normalized data from the plurality of EDR tools. A plurality of alerts are then generated from the normalized and enriched data along one or more alert pathways to improve fidelity of the plurality of alerts. The plurality of alerts are correlated with the contextual information of the plurality of assets as well as information from other data sources to identify a set of incidents of suspicious activities.

The invention discloses an intelligent video playing control system based on multi-mode interaction, and the system comprises four core modules: a voice instruction collection and preprocessing module which collects voice through a microphone array and extracts clear signals through adaptive filtering noise reduction and acoustic model endpoint detection; the multistage speech recognition and intention analysis module is used for converting speech into text by using a deep learning model, judging an instruction category in combination with an intention classification model, and extracting key parameters through a named entity recognition model; the context perception and instruction optimization module is used for maintaining a video playing state in real time based on a state machine model and eliminating instruction ambiguity in combination with a historical instruction; and the precise video playing control module calls a player API (Application Program Interface) to realize precise operations such as frame-level fast forward / fast backward, multi-speed adjustment, playing / pause switching and playing exit. The voice control accuracy and the control refinement degree can be remarkably improved, dependence on entity operation components is not needed, and the method is suitable for various devices such as mobile phones, televisions and vehicle-mounted systems.

This disclosure describes techniques for mapping local device identifiers used in monitoring data from different sources to a common global identifier to enable correlation of monitoring events related to the same device. The techniques can be used in the context of an Extended Detection and Response (XDR) system architecture for advanced threat detection and response in a computer system. In some cases, the XDR system ingests security data from various monitoring components like Endpoint Detection and Response (EDR), Intrusion Detection Systems (IDSs), Intrusion Prevention Systems (IPSs), firewall engines, and email security systems.

The present application relates to the technical field of semantic analysis, and particularly relates to a special operation data processing method for dangerous chemical product enterprises based on semantic analysis, comprising: endpoint detection is performed on a field voice stream to obtain a voice segment, and then a recognized text is obtained, and the recognized text is matched with a preset hot word to extract a recognized text feature; semantic analysis is performed based on the recognized text feature, an intention triple is extracted, and a semantic analysis confidence is determined; an individual risk basic value is calculated according to the intention triple and the semantic analysis confidence, combined with an individual historical risk tendency; a group risk feature vector and an operation space topology structure feature vector are extracted; a risk coefficient is determined by fusing the individual risk basic value, the group risk feature vector and the operation space topology structure feature vector, and hierarchical early warning is performed. The present application realizes accurate identification and hierarchical early warning of personnel risk in a restricted space fire-crossing operation scene by fusing multi-dimensional features.

The invention relates to a portal crane luffing mechanism end point detection and protection integrated device, and relates to the technical field of portal crane equipment, the portal crane luffing mechanism end point detection and protection integrated device comprises an induction piece arranged on a cantilever crane, and a first limit switch and a second limit switch which are arranged on a rotating seat, the first limit switch is arranged at the tail end of an amplification process, and the second limit switch is arranged at the tail end of an amplification process; the second limit switch is arranged at the tail end of the amplitude reduction process; a third limiting switch located behind the first limiting switch and a fourth limiting switch located behind the second limiting switch are arranged on the rotating base, and the first limiting switch and the third limiting switch can control on-off of the amplitude variation driving source piece in the amplitude increasing process. The second limit switch and the fourth limit switch can control the on-off of the amplitude variation driving source part in the amplitude reduction process respectively, and the induction part can trigger the first limit switch, the second limit switch, the third limit switch and the fourth limit switch. The method has the effect of improving the terminal point detection limiting reliability of the luffing mechanism.

The invention discloses a voice endpoint detection method, a related device, equipment and a medium, and the method comprises the steps: carrying out the detection of a to-be-detected audio based on an endpoint detection model, and obtaining a probability value that an audio frame in the to-be-detected audio belongs to a target frame type; wherein the target frame type is at least one of a plurality of preset frame types, the plurality of preset frame types comprise human voice and non-human voice, the endpoint detection model is obtained based on training of a first target audio, and a first target frame in the first target audio is classified into any one of a plurality of sample sets based on alignment information and a prediction probability; the plurality of sample sets comprise human voice difficult samples, non-human voice difficult samples, human voice easy samples and non-human voice easy samples, and training loss of the endpoint detection model comprises sub-loss of each sample set; and determining a final frame type of the audio frame based on the probability value that the audio frame belongs to the target frame type. According to the scheme, the accuracy of voice endpoint detection in a complex environment can be improved.