Abstract: The application provides an image detection method. A category label library may be pre-constructed based on the characteristics of images of a target type, and for each category label in the category label library, a probability threshold for that category label may be determined based on an accuracy rate of prediction results of a target detection model on sample images. By pre-setting, for each category label in the category label library, a probability threshold adapted to that category label, the target detection model achieves higher accuracy when detecting images of the target type, thereby reducing the occurrence of false detections or missed detections. Moreover, the probability threshold corresponding to each category label can be adjusted based on the accuracy rate of the prediction results of the target detection model, so that users can adjust the strictness of the detection criteria based on actual needs, making image detection more flexible.

Abstract: Methods and systems for imaging and analysis are described. Accurate pressure ulcer measurement is critical in assessing the effectiveness of treatment. However, the traditional measuring process is subjective. Each health care provider may measure the same wound differently, especially related to the depth of the wound. Even the same health care provider may obtain inconsistent measurements when measuring the same wound at different times. Also, the measuring process requires frequent contact with the wound, which increases risk of contamination or infection and can be uncomfortable for the patient. The present application describes a new automatic pressure ulcer monitoring system (PrUMS), which uses a tablet connected to a 3D scanner, to provide an objective, consistent, non-contact measurement method. The present disclosure combines color segmentation on 2D images and 3D surface gradients to automatically segment the wound region for advanced wound measurements.

Abstract: A model training method, a model deployment method, an electronic device, and a storage medium are provided, which relate to a field of artificial intelligence technology, in particular to fields of large model technology and deep learning technology. The model training method includes determining at least one module scale information according to preset model scale information; determining at least one to-be-trained module according to the at least one module scale information and a plurality of initial modules of an initial model, where the to-be-trained module includes a to-be-trained parameter determined from the initial module according to the module scale information; and training the at least one to-be-trained module according to training sample data to obtain at least one trained module, where the training sample data is obtained from at least one of text sample data, image sample data, audio sample data, or video sample data.

Abstract: Implementations for latent space broadening using lookalike thresholds for active learning strategies are provided. One implementation provides a computing system comprising: processing circuitry and memory storing instructions that, when executed, cause the processing circuitry to: receive a plurality of unlabeled data samples; determine a plurality of seed cases; for each of the seed cases, identify a subset of the unlabeled data samples that is semantically similar to the seed case; filter the identified subsets of the unlabeled data samples to remove data samples below a predetermined lookalike threshold, wherein the predetermined lookalike threshold describes a consistency level between an annotation and a prediction of an output of the machine learning model; annotate the filtered subsets of the unlabeled data samples to generate a labeled dataset; and train the machine learning model using the labeled dataset.

Abstract: A search method based on a neural network model is provided. The neural network model includes a semantic representation model, a recall model, and a ranking model. The present disclosure relates to the field of artificial intelligence, and in particular to the technical field of search. An implementation of the method comprises: inputting a target search and a plurality of objects to be matched to the semantic representation model to obtain a first output of the semantic representation model; inputting the first output of the semantic representation model to the recall model, and obtaining at least one recall object matching the target search from the plurality of objects to be matched by using the recall model; and inputting a second output of the semantic representation model to the ranking model, and obtaining a matching value of each of the at least one recall object by using the ranking model.

Abstract: Implementations for sample broadening techniques using large language model-based filtering to expand active learning strategies are provided. One implementation provides a computing system comprising: processing circuitry and memory storing instructions that, when executed, cause the processing circuitry to: receive a plurality of unlabeled data samples; determine a plurality of seed cases; for each of the seed cases, identify a subset of the unlabeled data samples that is semantically similar to the seed case; identify one or more of the unlabeled data samples having a prediction score difference above a predetermined threshold using a large language model; annotate the subsets of the unlabeled data samples and the one or more of the unlabeled data samples to generate a labeled dataset; and train the machine learning model using the labeled dataset.

Abstract: A device and method for measuring at least one cellular activity. The device includes: (a) a deformable polymeric base membrane having a first side and a second side, the polymeric base including a well having an opening on the first side, a cavity extending from the opening and a floor formed by the second side of the polymeric base membrane, (b) a deformable polymeric top membrane overlapping the second side of the polymeric base membrane; and (c) a sensing element disposed between the polymeric base membrane and the polymeric top membrane, the sensing element being disposed over the floor of the well, such that a portion of the second side that forms the floor of the well, the sensing element and a portion of the top membrane that overlaps the well form a suspended membrane.

Abstract: Some embodiments of the present invention provide a heat exchanger flat tube, a heat exchanger flat tube assembly and a heat exchanger. The heat exchanger flat tube includes: a flat tube body having a fin mounting section and a finless section which are communicated with each other, the finless section comprising a straight tube section and a bent section, the fin mounting section being communicated with the straight tube section through the bent section, and the fin mounting section and the straight tube section being oppositely arranged, the end of the fin mounting section away from the bent section is configured to be connected to a header. The present disclosure effectively solves the problem in the art known to inventors that the welding joint between a heat exchanger flat tube and a header is susceptible to corrosion damage.

Abstract: The present disclosure discloses a base editor and the use thereof. The present disclosure provides a nucleic acid base editor, specifically a base editor which is not based on CRISPR technology. The base editor comprises a sequence-specific DNA binding protein, a nickase, an exonuclease and a base-specific deaminase. This base editor is single-strand-specific, and as compared with conventional base editors, the base editor of the present disclosure has wide applicability in cells and is capable of functioning in the nucleus as well as in mitochondrial DNA and/or chloroplast DNA. This base editor has the characteristics of achieving base editing products with high purity and resulting in few indel byproducts while realizing efficient base editing, which is conducive to being used as an efficient and safe gene editing tool.

Abstract: The present disclosure relates to a multimedia presentation method, apparatus, readable medium and electronic device. The method comprises: in response to a user's rotation operation on a terminal, obtaining a rotation parameter of the terminal, the terminal presenting first multimedia data through a current page; in response to determining that the terminal rotates to a predetermined direction according to the rotation parameter, obtaining second multimedia data corresponding to the first multimedia data; switching a current page of the terminal to a target page and presenting the second multimedia data through the target page. In this way, without a click operation, it is able to switch to the target page through the rotation operation of the user, and present the second multimedia data corresponding to the first multimedia data, which improves the convenience and interest of the user's operation, thereby improving the user experience.

Abstract: Disclosed is a few-shot image classification method based on a hierarchical learning genetic programming algorithm: constructing a few-shot image classification system based on the hierarchical learning genetic programming algorithm; dividing the image data set into a training set and a test set; performing image preprocessing and feature extraction operations on the training set to construct a feature storage table; constructing an ensemble solution by using features in the feature storage table as terminal inputs, and optimizing a final classification effect by using an ensemble strategy based on an individual difference value; using the test set as an input of an image classification solution, outputting predicted class labels of the test set, and evaluating a performance of the image classification solution according to an actual label of the test set; using to-be-classified image data as the input of the image classification solution, and outputting an image classification result.

Abstract: Disclosed is a few-shot image classification method based on a hierarchical learning genetic programming algorithm: constructing a few-shot image classification system based on the hierarchical learning genetic programming algorithm; dividing the image data set into a training set and a test set; performing image preprocessing and feature extraction operations on the training set to construct a feature storage table; constructing an ensemble solution by using features in the feature storage table as terminal inputs, and optimizing a final classification effect by using an ensemble strategy based on an individual difference value; using the test set as an input of an image classification solution, outputting predicted class labels of the test set, and evaluating a performance of the image classification solution according to an actual label of the test set; using to-be-classified image data as the input of the image classification solution, and outputting an image classification result.

Abstract: Systems and methods are provided for estimating subject pose and shape. In some examples, the method includes generating, by an image encoder, a structured representation for an image; mapping, by a prompt encoder, a plurality of types of prompts associated with the image to a plurality of prompt tokens; and estimating, by using a transformer-based computing model, a set of subject parameters based at least in part on the structured representation and the plurality of prompt tokens. In some examples, the plurality of types of prompts includes a first type of prompts having a first modality and a second type of prompts having a second modality different from the first modality.

Abstract: Methods, computer program products, and systems are presented. The method computer program products, and systems can include, for instance: training a machine learning model with use of supervised learning, wherein the training includes applying to the machine learning model training data that includes historical logging data and historical metrics data of a computer environment, the training data being labeled with anomaly label data; inferencing the machine learning model for output of a data collection profile; processing current logging data and metrics data of the computer environment in dependence on one or more attribute of the data collection profile; identifying an anomaly of the computer environment in dependence on the processing; and performing an action for remediation of the anomaly.

Abstract: The present disclosure relates to the technical field of video playback, and provides a multimedia information playback method and apparatus, an electronic device, and a computer readable storage medium. The method includes: during a process of an information flow being browsed, in response to that display time of a target information entry in the information flow on a display interface of the information flow reaches a first time threshold, displaying a preset prompt message on an upper layer of the target information entry, so as to prompt a user to input a corresponding gesture; and in response to that a preset gesture operation to the target information entry inputted by the user is detected, playing multimedia information corresponding to the target information entry.

Abstract: A method for classifying environmental image frames for vehicles, using a universal iterative classification model, is presented. The method combines text-based querying, active machine-learning models, and user input to form an end-to-end automatic flow for sourcing video frames captured by sensors on a vehicle. An index of frames is used and continuously populated with data for new frames, with each frame scored on its likelihood of containing a representation of a driving scenario of interest. Each iteration of the classification model produces a classification result that is predicted to belong to the scenario of interest. Binary labels can be applied to the results. Subsequent iterative training of classification models can be performed using updated training sets containing previously labeled classification results, to improve precision and accuracy in classifying image data to a driving scenario.

Abstract: Disclosed in the present invention is a nanocrystalline soft magnetic alloy with high magnetic induction and high frequency. The nanocrystalline soft magnetic alloy has a molecular formula of FeaSibBcMadCuePf, in which M includes one or more of Nb, Mo, V, Mn, and Cr, molar percent contents of elements are as follows: 6?b?15, 5?c?12, 0.5?d?3, 0.5?e?1.5, and 0.5?f?3, and the balance includes Fe and impurities. A difference between an induced anisotropy value and an average magnetocrystalline anisotropy value is 0.1-1 J/m3. The soft magnetic alloy has high magnetic permeability and low magnetic loss at high frequency. Further disclosed in the present disclosure is a method for preparing the nanocrystalline soft magnetic alloy with high magnetic induction and high frequency.

Abstract: The present disclosure provides Pyruvate Dehydrogenase Kinase 4 (PDK4) as a target for intervention of cellular senescence and a use in chemotherapy against cancer. The present disclosure reveals PDK4 as a novel target for intervention of cellular senescence and intervention in drug resistance of tumors. By targeting PDK4, drugs can be developed to inhibit drug resistance of tumors, suppress or eliminate the SASP. Additionally, drugs targeting PDK4 can be used in combination with chemotherapeutic drugs for diagnosis or prognosis of related diseases or symptoms based on this target and mechanism.

Abstract: An image registration method using a fast differential evolution sample consensus algorithm based on historical records: 1: image acquisition: acquiring a to-be-registered image and a reference image; 2: feature extraction: respectively extracting point features of the to-be-registered image and the reference image by using a SIFT algorithm; 3: feature matching: performing coarse feature matching between the to-be-registered point feature set and the reference point feature set to obtain a sample set; 4: parameter setting; 5: sorting the sample set in ascending order, and initializing a sample inlier archive AS and a model inlier archive AM; 6: first-time initialization of a population; 7: updating a fast test execution threshold uT; 8: mutation phase; 9: selection phase; 10: mutation operator parameter update; 11: resident initialization of the population; 12: repeating 7-11; 13: selecting the best individual from the population to perform spatial transformation on the to-be-registered image to obtain a regi

Abstract: The present disclosure discloses a base editor and the use thereof. The present disclosure provides a nucleic acid base editor, specifically a base editor which is not based on CRISPR technology. The base editor comprises a sequence-specific DNA binding protein, a nickase, an exonuclease and a base-specific deaminase. This base editor is single-strand-specific, and as compared with conventional base editors, the base editor of the present disclosure has wide applicability in cells and is capable of functioning in the nucleus as well as in mitochondrial DNA and/or chloroplast DNA. This base editor has the characteristics of achieving base editing products with high purity and resulting in few indel byproducts while realizing efficient base editing, which is conducive to being used as an efficient and safe gene editing tool.