120results about How to "Reduce false positives" patented technology

An intrusion detection system (IDS) is capable of identifying the source of traffic, filtering the traffic to classify it as either safe or suspect and then applying sophisticated detection techniques such as stateful pattern recognition, protocol parsing, heuristic detection or anomaly detection either singularly or in combination based on the traffic type. In a network environment, each traffic source is provided with at least one IDS sensor that is dedicated to monitoring a specific type of traffic such as RPC, HTTP, SMTP, DNS, or others. Traffic from each traffic source is filtered to remove known safe traffic to improve efficiency and increase accuracy by keeping each IDS sensor focused on a specific traffic type.

A system and method for providing distributed security of a network. Several device profilers are placed at different locations of a network to assess vulnerabilities from different perspectives. The device profiler identifies the hosts on the network, and characteristics such as operating system and applications running on the hosts. The device profiler traverses a vulnerability tree having nodes representative of characteristics of the hosts, each node having an associated set of potential vulnerabilities. Verification rules can verify the potential vulnerabilities. A centralized correlation server, at a centrally accessible location in the network, stores the determined vulnerabilities of the network and associates the determined vulnerabilities with attach signatures. Traffic monitors access the attack signatures and monitor network traffic for attacks against the determined vulnerabilities.

Systems and methods to reduce false positives in spam filtering. In one aspect, a method includes automatically generating profiles for persons identified in messages, such as incoming and outgoing emails. Messages flagged as spam can be checked against the profile to identify false positives.

The invention discloses a video monitoring method and a video monitoring system. The method comprises the following steps: S100, acquiring a video stream to be analyzed currently; S200, detecting whether a motion object exists in the video stream, if yes, carrying out background modeling, morphological process and eliminating shadow, if no, carrying out circular detection; S300, using a particle filtering or a Kalman filtering tracking method to track the motion object; S400, analyzing the shape information and motion characteristics of the motion object, analyzing abnormal situations by utilizing dispersion degree and area information, classifying the motion object and accomplishing abnormal action identification process analysis; and S500, making an emergency alarm when abnormal situation happens to the video stream. With the method in the invention, the motion object can be analyzed intelligently, multiple bottom layer calculation for one same video stream is carried out only once, and miss report and false report rate of abnormal action in monitoring can be reduced; and with the method. Besides, the method also can be linkaged with security, so that monitoring personnel can make response in time.

The invention relates to a method and device for adjusting target hitting characteristics according to attacking logs. An IP address of a terminal can hit the plurality of target hitting characteristics in a certain period of time when the terminal attacks a protected web server, and each target hitting characteristic corresponds to a small number of attacking logs. Therefore, the target hitting characteristics and target websites of all attacking logs in the same IP address in the first time interval are the same, and when the number of all the attacked logs in the first time interval is larger than or equal to a first preset threshold, it is considered that a false alarm is given when the IP address browses the target website, the IP address and the target website are simultaneously avoided in the target hitting characteristics, and therefore the terminal with the IP address is not intercepted when the terminal browses the target website and is matched with the target hitting characteristics, terminals with other IP addresses are intercepted when the terminals browse the target website and are matched with the target hitting characteristics, and the attacking logs are generated. False alarms are reduced without directly abolishing a certain rule and on the premise of not reducing the defending effect.

The invention provides a method and system for calculating a passenger crowdedness degree, comprising steps of establishing a video data collection environment, collecting video data of passengers getting on and off, reading collected video data of passengers getting on and off, performing continuous multi-frame pre-processing on the video data, identifying people heads according to the pre-processing result, taking the detected human head as a target object to be tracked by mean value drift, determining the behaviors of getting on and off of the passengers in the area where the target object is positioned and determining the crowdedness degree of the passengers inside the vehicle according to the numbers of the passengers getting on and off. The invention also provides a system for calculating the passenger crowdedness degree. The technical scheme of the invention adopts the pre-processing which can eliminate the static background to overcome the interference on the identification of the head top of the people in the frame image which is caused by the change of the light, and can effectively reduce the false detection, leak detection and mistake detection through limiting the size of the detection window.

The invention discloses a method and device for known variation detection based on Bayes testing and Poisson's distribution testing. The method comprises the steps that a sequencing and length-reading base sequence, a reference genomic sequence and a sequencing and length-reading sequence reckoned under assumption of existence of a known variance point are provided; comparative detection is conducted on the sequencing and length-reading sequence reckoned under the assumption of existence of the known variance point and the sequencing and length-reading base sequence; and a Bayes testing model and a Poisson's distribution testing model are judged. The method disclosed by the invention has the advantages that SNV or InDel variance can be detected in a high-specificity and high-sensitivity manner.

The invention discloses a remote sensing image change detection method based on a twin convolutional neural network, relates to the field of remote sensing, and mainly solves the problem of poor generalization of a conventional change detection method at present. The method comprises the following steps: acquiring multi-temporal remote sensing image data; obtaining mask images, establishing a remote sensing image change detection data set; constructing a twin convolutional neural network model; training a twin convolutional neural network by using the data set; obtaining a training model, carrying out change detection on a front time phase image and a rear time phase image to be detected by utilizing the training model; and obtaining a preliminary change prediction result, and comparing the prediction value of the pixel of the preliminary change prediction result with a preset pixel threshold, thereby dividing the preliminary change prediction result into a change region category and anon-change region category, and obtaining a change detection result. The method is better in generalization performance, meanwhile, end-to-end processing is met, and engineering application is facilitated.

The invention relates to a fatigue driving early warning wristband which integrates heart rate and driving action, and belongs to the technical field of smart wristbands and driving assistance systems. The heart rate sensor on the back of the bracelet, the heart rate sensor module embedded in the bracelet shell, the three-axis acceleration sensor module and the microprocessor chip. The method is to measure the heart rate of the human body and output real-time heart rate signals; detect the action acceleration of the driver's hand in the three dimensions of X, Y, and Z, and output real-time three-dimensional action acceleration information; Comprehensive judgment of fatigue driving situation. If it is judged that the current driving state is fatigued, the micro vibration motor will be driven to generate vibrations for fatigue warning. The bracelet of the present invention fuses and analyzes the detected two types of information, thereby realizing accurate judgment and early warning of fatigue driving.

The invention provides a novel improved immunochromatographic test strip, and preparation and application thereof. According to the invention, a buffer pad is additionally arranged between a reaction membrane and a colloidal particle conjugate releasing pad with a marked antibody against a to-be-detected object, so colloidal particles of the antibody against the to-be-detected object disintegrated from the releasing pad can be more uniformly dissolved in a detection sample in the buffer pad before falling onto the reaction membrane, adequate reaction time is obtained, and a corresponding target detection object undergoes more complete immunoreaction; thus, the reaction efficiency of the antibody against the to-be-detected object and the target detection object in the sample is maximized. Compared with conventional immunity immunochromatographic test strips, the novel improved immunochromatographic test strip in the invention has the advantages of high sensitivity, low false positive diagnosis rate, uniform color of a test line; and a preparation method for the novel improved immunochromatographic test strip low in cost, simple to operate and applicable to preparation of immunochromatographic test strips of a variety of forms.

The invention provides a network malicious data detection method, device and system, and the method comprises the steps: firstly collecting network data, and extracting a data characteristic value in the network data; secondly obtaining the correlation between the data characteristic value and a malicious data characteristic value in a pre-stored malicious characteristic set; and finally carrying out the expansion of the malicious data characteristic set according to the data characteristic value when the correlation between one data characteristic value and the malicious data characteristic value exceeds a preset correlation threshold value. The method can continuously correct and optimize the malicious data characteristic set, achieves the prediction judgment of unknown network threats and the active defense of the network safety, and reduces the false alarm and alarm leakage rates of intrusion detection of network malicious data.

The invention provides an accurate and reliable method and an accurate and reliable system for treating medical images and aims to solve a problem in the prior art. The method comprises steps that A, multiple original sample medical images after lesion point calibration are acquired; B, data pre-treatment on the multiple original sample medical images is carried out to acquire multiple training sample medical images; C, depth neural training for the multiple training sample medical images is carried out to acquire a lesion point identification model; and D, test medical images are inputted to the lesion point identification model to acquire a lesion point identification result.

The invention relates to a rule based interactive Chinese spam filtering method. The method comprises a first step of extracting a Chinese spam filtering rule from Chinese spam samples; a second step of training by use of an improved genetic algorithm to obtain an optimum weight distribution of the rule; a third step of intensifying interaction between a user and a server, collecting and analyzing false negative and false positive information that are fed back by the user; and a fourth step of regulating the weight of a sensitive rule in time through an incremental learning method. In the method, the rule weight distribution is trained by use of the improved genetic algorithm, the false negative and false positive information is collected through interaction between the user and the server, and the rule weight is corrected by use of the incremental learning method according to an analysis result, so that false positive rate and false negative rate are lowered, and the filtering adaptivity is improved. The interactive filtering method reduces misjudgement and has a better filtering effect while guaranteeing that identification rate of spam is not lowered. The incremental learning method lowers the retraining frequency, makes weight updating faster, and improves the overall performance of a filtering system.

The invention discloses an article detection method and apparatus. The invention aims to solve the problem of low accuracy of human eye-based identification which is caused by misdetection and misseddetection caused by the fatigue of human eyes. According to the embodiments of the present invention, the method includes the following steps that: a plurality of different neural networks are adoptedto extract article features in a target image, so that feature images corresponding to each neural network are obtained; the feature images corresponding to each neural network are subjected to feature fusion, so that a fused image can be obtained; and the fused image is matched with an image corresponding to a preset item, and an item included in the target image is determined according to a matching result. Since the neural networks are adopted to realize the method provided by the embodiments, manual inspection is avoided, and therefore, the phenomena of misdetection and missed detection are reduced, and accuracy is improved.

The invention provides a family anti-theft monitoring device achieved through the internet, and belongs to the fields such as the electronics field and the communication field. The family anti-theft monitoring device achieved through the internet comprises a camera shooting input module, a door magnet and window magnet module, a human body infrared induction module, a sound and vibration induction module, an internet communication module, an embedded type core processor, a wireless gateway module, a memorizer, a deploying and controlling alarming starting/relieving module, an identification recognition module, an internet central computer, a network computer and a mobile telephone. The embedded type core processor recognizes a moving object according to image information input by a camera, door magnets, window magnets and infrared sensors can also recognize an illegal intruder, and a user can monitor the dynamic image information through the internet.

The invention discloses a driving behavior sensing method based on information fusion and system thereof. The method includes the following steps of collecting acceleration, vehicle speed and course angle of a vehicle by means of a sensor assembly integrated on the vehicle; according to the acceleration, the vehicle speed and the course angle, conducting horizontal alignment and course alignment on a vehicle coordinate system and detecting the alignment results; according to the acceleration, the vehicle speed and the course angle, judging whether or not behaviors such as speed increase, speedreduction, steering and collision happen to the vehicle, and recording and uploading related data if yes. The acceleration, the vehicle speed, the course angle and other different kinds of vehicle data of the vehicle are obtained through an acceleration meter (namely a gyroscope or a compass), a GPS module and the like. The acceleration, angular velocity, a magnetic direction and other multidimensional data such as the course angle, the height and the vehicle speed in a GPS can be utilized for fusion analysis, the probability of misdetection and measurement omissions is substantially decreased, and complete data during occurrence of an event can be accurately recorded to provide bases for further analysis.

The invention relates to an automatic human body physiological parameter acquisition system and an automatic human body physiological parameter acquisition method. The system consists of at least one physiological parameter detection device and at least one identity recognition ID card which is matched with the physiological parameter detection device, wherein each physiological parameter detection device mainly comprise a main control circuit module, an NFC near field communication module and a sensor interface module, and the main control circuit module is provided with a display module, a keyboard, a wireless communication module, a printing module, a microprocessor circuit and a power supply management module. The main control circuit module is connected with one or the combination of any of a blood oxygen level sensor, an electrocardio sensor, a blood pressure sensor, a pulse sensor and a respiratory sensor through the connected physiological parameter sensor interface module to acquire the human body physiological parameter data and also recognizes and reads a user identity recognition ID card brought by the specific human body through the connected NFC near field communication module.

The invention discloses a contact net pantograph stagger value overrun defect recognition method and system.The method comprises the steps that an image processing technology is adopted to calculate a stagger value of pantograph, condition limiting is conducted on stagger value overrun judgment according to factors such as stagger value smoothness, whether in an extreme value state or not and a contact line mode, and alarm interference is removed.According to the contact net pantograph stagger value overrun defect recognition method and system, the problem of misinformation of a stagger value overrun defect is effectively solved, stagger value overrun judgment can be accurately conducted, and the pantograph abnormal condition checking method is efficient, convenient and rapid.

The invention discloses a lightweight DDoS attack detection device and method on a high-speed network. The device comprises a CBFSketch module and a DDoS attack detection module. The method comprisesthe following steps: S1, acquiring network flow, and recording and storing data flow information; S2, according to the recorded data flow information, inquiring the flow size or occurrence frequency of a specific data flow, and judging whether the data flow is DDoS attack flow or not according to an inquiry result; and S3, setting a threshold value, detecting the data flow exceeding the thresholdvalue according to the recorded data flow information, and completing DDoS attack detection. The multi-level Sketch structure is adopted to store the data flow, the space can be dynamically applied, the space overhead upper limit can be effectively controlled, the flow exceeding the threshold value can be rapidly detected to detect the DDoS attack, misinformation is reduced, and the detection accuracy is improved.

The invention provides an operation state monitoring method of a railway through line cable and belongs to the field of on-line monitoring of the railway through line cable. The operation state monitoring method comprises a high-speed railway through line cable and is characterized in that by detecting a bus PT zero-sequence voltage transient-state signal of the railway through line cable is generated or not, whether an instantaneity transient-stat fault occurs to the high-speed railway through line cable is judged; and by monitoring and recording electromagnetic transient-state signal which is generated by the high-speed railway through line cable and flows through a metal shielding layer, fault early warning is performed on the high-speed railway through line cable and the fault point distance is indicated. Compared with the original cable operation state monitoring method, the operation state monitoring method provided by the invention is advantageous in that early warning can be given to an instantaneity transient-stat grounding fault before a cable is damaged; non-fault disturbance signals generated by passing of a train, reclosing and the like can be filtered; system misinformation caused by disturbance is effectively reduced; alarm accuracy is improved; and false alarm rate is reduced.

The invention discloses a battery thermal runaway early warning method and device. The battery thermal runaway early warning method comprises the steps of: S1, calculating the leakage current of eachsingle battery according to the statistical data in a preset time range; S2, collecting battery thermal runaway characteristic data in real time through a battery management system BMS, and acquiringa real-time diagnosis result indicating whether thermal runaway occurs or not according to the battery thermal runaway characteristic data; and S3, judging whether thermal runaway occurs or not according to the size relationship between the leakage current of each single battery and a diagnosis threshold or according to the real-time diagnosis result, and carrying out early warning when it is judged that thermal runaway occurs. The battery thermal runaway early warning method is close to the actual demand for the thermal runaway early warning function at present, reduces the false alarm and missing alarm probability of thermal runaway early warning, reserves as much time as possible for user escape and after-sales rescue, and greatly improves the reliability of battery thermal runaway early warning.

The invention belongs to the fire alarm detection technology field and provides a redundant smog detector and a system. The redundant smog detector comprises controllers connected with a power source portion, a code receiving portion and a code returning portion, wherein the controllers comprise a first controller and a second controller, the first controller is connected with a first smog sensing portion, and the second controller is connected with a second smog sensing portion; for the system, the redundant smog detector is in hitched with a bus, the bus is connected with a first general controller and a second general controller, the first general controller comprises a first system control portion and a first code transceiving portion, the second general controller comprises a second system control portion and a second code transceiving portion, and the first code transceiving portion and the second code transceiving portion are respectively connected with the bus. According to the redundant smog detector and the system, problems that multiple complex conditions can not be satisfied by a smog detector in the prior art, a damage rate is high, service life is short, error reports constantly occur, detection precision seriously decreases as time goes by, and the system facing various situations can not work normally are solved.

The invention discloses a tramcar automatic ticket selling and checking inbound gate. The part of a gate box, that is arranged the inbound inlet of a gate, is provided with a coin inputting slot. The part of the gate box, that is arranged right below the coin inputting slot, is provided with a coin inputting channel. A coin identification device is arranged inside the coin inputting channel. A coin withdrawing channel and a cashbox channel are arranged below the coin inputting channel. The coin identification device arranged inside the coin inputting channel is used for identifying true and false coins. If coins are identified to be true by means of the coin identification device, the true coins are driven to enter the cashbox channel. Meanwhile, an industrial personal computer controls a three-rod mechanism to rotate so as to enable passengers to pass by. The coin inputting slot is provided with a waterproof cover. According to the technical scheme of the invention, the tramcar automatic ticket selling and checking inbound gate is applicable to tramcars and integrates the functions of card swiping and coin inserting as a whole. The gate box is provided with a vibration sensor, thus having the functions of alarming and anti-stealing. The coin inputting slot is provided with the waterproof cover. Therefore, the rainwater is prevented from flowing into the gate box via the coin inputting slot.

The invention relates to a distributed optical fiber temperature measuring device for a normal temperature conveying pipeline. The device comprises a conveying pipeline, a temperature sensing opticalcable, a heater and an outer pipeline; the conveying pipeline is used for conveying a normal temperature medium; the temperature sensing optical cable and the heater are laid in parallel with the conveying pipeline and are located below the conveying pipeline; the outer pipeline is used for sealing the conveying pipeline, the temperature sensing optical cable and the heater in a closed space; thetemperature sensing optical cable is connected with a DTS host; and the heater is a passive heat source. According to the distributed optical fiber temperature measuring device for the normal temperature conveying pipeline of the invention, the external pipeline can accumulate liquid, and the heater can perform quick heating, and therefore, faster and more accurate leak detection can be realized,and the possibility of false alarms can be greatly reduced.

The invention relates to the technical field of escalator safety monitoring, and discloses an escalator retention detection method and system based on a Gaussian mixture model. The method comprises the steps: A), extracting pedestrian information from a video image through a deep learning network; B) constructing a classification model by using a machine learning method, and classifying and screening pedestrian information detected in real time; C) constructing a Gaussian mixture model, and obtaining a retention target image of the current frame; D) preprocessing the detention target image toobtain an image in the detention detection area at the entrance and exit of the escalator; E) acquiring a candidate retention target contour by adopting an edge detection algorithm, and judging the area of a candidate retention box; and F) carrying out a candidate retention frame overlapping judgment, and performing retention alarms. Detection and real-time alarm are carried out on retention of articles and passengers at the entrance and exit of the escalator, the purpose of monitoring retention behaviors of the entrance and exit area of the escalator is achieved, and the retention detection accuracy is high.

The invention discloses an active defense method for power industry network security. The method comprises the steps of obtaining and analyzing behaviors of access users; and defining and explaining behaviors of the access user according to an analysis result, and performing active defense by adopting a firewall, a honeypot and a WAF. The invention provides an active defense method for power industry network security. An existing defense strategy is utilized, meanwhile, a spoofing defense method is creatively added, safety three-dimensional defense of the whole power business system is carriedout under regulation and control of the unified defense strategy, and the protection capacity of the network environment of the power industry is improved to the maximum extent. Moreover, the methodcan construct a system capable of quickly detecting, reducing missing reports and immediately blocking malicious behaviors, and is high in protection capability, diversified in protection means and relatively good in protection effect.

The invention relates to the field of biotechnology, in particular to an enzyme-linked immunosorbent assay method for an enterovirus 71 (EV71 for short)-specific antibody in serum. Through a method of indirect enzyme-linked immunosorbent assay (ELISA), a prepared EV71 viral antigen is immobilized on an ELISA plate, and the antibody is detected through specific adsorption of antigen and antibody. Since the coated antigen comprises a specific antigen epitope for the EV71 virus, the method can be used for detecting specific antibody for the EV71 virus in the serum, and false positive of serum detection is reduced. The enzyme-linked immunosorbent assay method for the enterovirus 71-specific antibody is suitable for detecting a large batch of animal or human serum samples, and is suitable for large-scale serological and epidemiological investigation.

The invention discloses a mobile phone detecting door. The mobile phone detecting door comprises a door body part and a data detecting part, wherein the data detecting part is arranged in a manner of being matched with the door body part. The mobile phone detecting door disclosed by the invention has the beneficial effects that the defects of high false alarm rate and low reliability and the like in the prior art can be overcome so as to realize the advantages of low false alarm rate and high reliability.

The invention discloses a transverse positioning method for a vibration source of a distributed optical fiber vibration sensor. The transverse positioning method comprises steps of S1, acquiring space-time vibration data of optical fiber spatial distribution caused by a single-point vibration source; S2, calculating the position of an optical fiber axial vibration center, closest to a vibration source, on the sensing optical fiber according to the space-time vibration data; S3, respectively acquiring vibration data wave widths delta t1 and delta t2 of the optical fiber axial vibration center and a monitoring point spaced from the optical fiber axial vibration center by a distance d through the vibration space-time data; and S4, obtaining a transverse distance between the vibration source and the axial vibration center of the optical fiber according to the seismic wave dispersion principle. The method is advantaged in that the transverse position information of the vibration source canbe obtained through a very small amount of calculation, and misinformation in practical application is greatly reduced.

The invention discloses a fan abnormal sound detection system and relates to the technical field of fan detection. The fan abnormal sound detection system comprises a sensor module, a signal collection module, a data processing module and a fixed support. The sensor module comprises an acceleration sensor, a sound sensor and a thermoelectric converter. The signal collection module comprises an NI9234 data collection card and an NI cDAQ-9188XT Ethernet cabinet. The NI9234 data collection card is connected with the acceleration sensor and the sound sensor and is further connected with the NI cDAQ-9188XT Ethernet cabinet. The data processing system comprises a data conversion and characteristic extraction module based on Labview and a mathematic modeling analysis system module based on Matlab. The data conversion and characteristic extraction module based on Labview is connected with the NI cDAQ-9188XT Ethernet cabinet and the mathematic modeling analysis system module based on Matlab. According to the invention, signal collection, analysis and judgment are integrated, and the manual misjudgement problems caused by inconsistent boundary judgment are reduced; in addition, the automatic degree is high, manual cost is saved, and the production efficiency is improved.