Abstract: In an example, a method for generating responses by a Machine Learning (ML) system includes processing, by a first language model, a natural language instruction to generate an instruction representation based on a meaning of the natural language instruction; translating, by a translation module comprising an interface between the first language model and a second language model, the instruction representation into data indicating an intent of the natural language instruction, wherein the second language model is trained with domain specific knowledge; providing, by the translation module, the natural language instruction and the data indicating the intent of the natural language instruction to the second language model; and generating, by the second language model, a response based on the natural language instruction and the data indicating the intent of the natural language instruction.

Abstract: In general, various aspects of the techniques are directed to causal analysis using large scale time series data. A computing system may convert large scale time series data to first time period records and second time period records according to a multi-scale time resolution. The computing system may implement a hierarchical machine learning model to generate embeddings that capture temporal characteristics of features of the large scale time series data. The computing system may generate a graph data structure indicating cause and effect correlations between features of the large scale time series data based on temporal dynamics captured in the cause and second time period records and/or the embeddings.

Abstract: The invention discloses an orthogonal dual-layer grating dynamic intensity modulation segmentation method based on quadrant, specifically include the following steps: S1: the fluence distribution under each beam is calculated through the radiation treatment planning system; S2: use orthogonal double-layer collimator for fluence segmentation; S3: divide the quadrant, divide the field surrounded by the upper, lower, left and right leaves into at least two quadrants, to obtain the fluence distribution and the corresponding leaf sequence of each quadrant; S4: perform regional planning of the fluence in each quadrant to obtain multiple different regions and determine the segmentation mode of different regions; S5: for any quadrant, use two mutually orthogonal leaf groups for segmentation.

Abstract: A computer-implemented method is provided that includes transmitting, by a master node to a plurality of computing nodes, definition information about an initial medical validation model (410): performing, by the master node, a federated learning process together with the plurality of computing nodes (420), to jointly train the initial medical validation model using respective processed local training datasets available at the plurality of computing nodes, the respective local training datasets being processed by the plurality of computing nodes based on the definition information; and determining, by the master node, a final medical validation model based on a result of the federated learning process (430). Through the solution, by means of federated learning, it addresses the data security and privacy concerns from local sites owning.

Abstract: An an energy storage apparatus is provided. the energy storage apparatus includes: a plurality of single batteries and a bipolar plate, each of the plurality of single batteries including a battery cell, the battery cell including a first electrode and a second electrode with opposite polarities and being insulated from each other; the bipolar plate is arranged between two adjacent single batteries, the first electrode or the second electrode of one of the adjacent single batteries forms an electrical connection with the bipolar plate through contacting, and the second electrode or the first electrode of the other of the adjacent single batteries forms an electrical connection with the bipolar plate through contacting.

Abstract: The invention provides a method and device for IMAT using orthogonal double layer multi leaves collimators. The method includes: discretizing the rotating arc into multiple equally spaced fields; using the conjugate gradient method to calculate the field intensity matrix; using the double-layer grating static segmentation algorithm to calculate the subfields of each field to obtain the first predetermined number of subfields with the largest contribution to the field of each field; selecting two subfields with similar shapes from the first predetermined number of subfields with the largest contribution, distributing them to the arc of rotation, and performing interpolation to obtain discrete subfields; calculating deposition Matrix; iterative calculation of the shape and weight of subfield; using Monte Carlo dose algorithm to calculate the intensity-modulated dose distribution.

Abstract: In an example, a system includes processing circuitry in communication with storage media. The processing circuitry is configured to execute a machine learning system including at least a first module, a second module and a third module. The machine learning system is configured to train one or more machine learning models. The first module is configured to generate augmented input data based on the streaming input data. The second module includes a machine learning model configured to perform a specific task based at least in part on the augmented input data. The third module configured to adapt a network architecture of the one or more machine learning models based on changes in the streaming input data.

Abstract: In general, techniques are described that address the limitations of existing conformal prediction methods for cascaded models. In an example, a method includes receiving a first validation data set for validating performance of an upstream model of the two or more cascaded models and receiving a second validation data set for validating performance of a downstream model of the two or more cascaded models wherein the second validation data set is different than the first validation set; estimating system-level errors caused by predictions of the upstream model based on the first validation data set; estimating system-level errors caused by predictions of the downstream model based on the second validation data set; and generating a prediction confidence interval that indicates a confidence for the system based on the system-level errors caused by predictions of the upstream model and based on the system-level errors caused by predictions of the downstream model.

Abstract: A method, apparatus, and system for determining an uncertainty estimation of at least one layer of a neural network includes identifying a neural network to be analyzed, representing values of each layer of the neural network as respective variable nodes in a graphical representation of the neural network, and modeling connections among each of the layers of the neural network as different respective factors across the variable nodes in the graphical representation, the graphical representation to be used to determine the uncertainty estimation of at least one layer of the neural network. The method, apparatus, and system can further include propagating data through the graphical representation to determine the uncertainty estimation of the neural network.

Abstract: This disclosure describes machine learning techniques for capturing human knowledge for performing a task. In one example, a video device obtains video data of a first user performing the task and one or more sensors generate sensor data during performance of the task. An audio device obtains audio data describing performance of the task. A computation engine applies a machine learning system to correlate the video data to the audio data and sensor data to identify portions of the video, sensor, and audio data that depict a same step of a plurality of steps for performing the task. The machine learning system further processes the correlated data to update a domain model defining performance of the task. A training unit applies the domain model to generate training information for performing the task. An output device outputs the training information for use in training a second user to perform the task.

Abstract: A compensation method for bed surface drop before and after accelerator radiotherapy bed includes the following contents: the weight G is decomposed into N weights on average, the length L is decomposed into M positions on average, and then an experiment is carried out. The drop matrix HN×M of the couch top is measured by the experiment, by performing cubic spline interpolation on the rows of the matrix HN×M, and then performing cubic spline interpolation on the columns, the resulting cubic spline interpolation matrix SN×N1, M×M1 of HN×M has N*N1 rows and M*M1 columns, and then applies it to practice. The drop amount H obtained by the compensation method for bed surface drop before and after accelerator radiotherapy bed of the present invention is the optimal estimated value with the Least Sum of Square Error.

Abstract: A method and system for real-time calculating a microseismic focal mechanism based on deep learning is provided, which belongs to the technical field of microseismic monitoring. The method includes: creating a training dataset, the training data including simulated DAS microseismic strain data and a focal mechanism corresponding to the simulated DAS microseismic strain data; training a focal mechanism calculation model by using the training dataset, with the simulated DAS microseismic strain data as an input and the focal mechanism corresponding to the simulated DAS microseismic strain data as a target output, so as to obtain a trained focal mechanism calculation model; collecting DAS microseismic strain data by a surface and downhole DAS acquisition system; performing preprocess operations such as removing abnormally large values on the DAS microseismic strain data; inputting the preprocessed DAS microseismic strain data into a trained focal mechanism calculation model to obtain a focal mechanism.

Abstract: The invention discloses a dynamic intensity-modulated segmentation method for an orthogonal double-layer grating blade device. The core of the segmentation algorithm is to construct a virtual single-layer grating after the velocities of the two-layer gratings are synthesized to perform dynamic intensity modulation of the single-layer grating (sliding-window) segmentation, and finally use two layers of gratings to conform to each segment. In order to reduce the segmentation error, the invention provides two optimization methods: blade motion trajectory optimization method and segment weight optimization method. The blade motion trajectory optimization method is to optimize the objective function under certain constraints with the motion trajectory of each blade as a variable under the condition that the segment weight is fixed. Segment weight optimization method is to optimize the time points of each segment when the blade motion trajectory is fixed.

Abstract: A method and system for real-time calculating a microseismic focal mechanism based on deep learning is provided, which belongs to the technical field of microseismic monitoring. The method includes: creating a training dataset, the training data including simulated DAS microseismic strain data and a focal mechanism corresponding to the simulated DAS microseismic strain data; training a focal mechanism calculation model by using the training dataset, with the simulated DAS microseismic strain data as an input and the focal mechanism corresponding to the simulated DAS microseismic strain data as a target output, so as to obtain a trained focal mechanism calculation model; collecting DAS microseismic strain data by a surface and downhole DAS acquisition system; performing preprocess operations such as removing abnormally large values on the DAS microseismic strain data; inputting the preprocessed DAS microseismic strain data into a trained focal mechanism calculation model to obtain a focal mechanism.

Abstract: The present disclosure discloses a high-safety ternary positive electrode material and a method for preparing the same; wherein the ternary positive electrode material has a chemical composition of Lia(NixCoyMn1-x-y)1-bMbO2-cAc, wherein 0.75?a?1.2, 0.75?x<1, 0<y?0.15, 1?x?y>0, 0?b?0.01, 0?c?0.2, M is one or more selected from the group consisting of Al, Zr, Ti, Y, Sr, W and Mg, and A is one or more selected from the group consisting of S, F and N; and CMn?(1?x?y)?0.07; CCo?y?0.05; 0?[CMn?(1?x?y)]/(CCo?y)?2.0. The ternary positive electrode material of the present disclosure is a high-nickel single crystal material with gradient concentration; it has the advantages of high capacity and high thermal stability, and the preparation method is simple, and is suitable for large-scale production.

Abstract: A method, apparatus and system for training an embedding space for content comprehension and response includes, for each layer of a hierarchical taxonomy having at least two layers including respective words resulting in layers of varying complexity, determining a set of words associated with a layer of the hierarchical taxonomy, determining a question answer pair based on a question generated using at least one word of the set of words and at least one content domain, determining a vector representation for the generated question and for content related to the at least one content domain of the question answer pair, and embedding the question vector representation and the content vector representations into a common embedding space where vector representations that are related, are closer in the embedding space than unrelated embedded vector representations. Requests for content can then be fulfilled using the trained, common embedding space.

Abstract: Embodiments of the present disclosure provide a DAS same-well monitoring real-time microseismic effective event identification method based on deep learning, including: constructing a DAS-based horizontal well microseismic monitoring system; constructing a training data set, including microseismic event data, pipe wave data and background noise data with different types of labels; constructing a signal identification module; training the signal identification module by using the training data set; preprocessing actual monitoring data, inputting the preprocessed data into the signal identification module to obtain an output result; marking microseismic events identified in the output result, and updating the marked microseismic events into the training data set; and adjusting and updating the signal identification module. The identification method according to the present disclosure can identify microseismic events in DAS same-well monitoring data in real time and efficiently.

Abstract: The invention disclosed a dynamic intensity modulation method and device based on orthogonal double-layer grating rotation sweep. The method comprises: 1) obtain the fluence intensity distribution of each beam through the radiotherapy planning system; 2) the beam field is preliminarily divided into four quadrants, which is surrounded by four groups of leaves from the top, bottom, left and right, each quadrant corresponds to the beam intensity distribution in a region within the beam field range, and corresponds to a pair of mutually orthogonal leaves; 3) for the beam intensity distribution in any quadrant, two groups of orthogonal leaves are used for segmentation; 4) synchronize the monitor unit MU of each quadrant; 5) obtain the motion trajectory of active leaves in each quadrant, the motion trajectory of passive leaf in each quadrant and the overall monitor unit MU by calculation.

Abstract: Disclosed is a lithium ion battery. The lithium ion battery comprises: a positive electrode piece, comprising a positive electrode coating area and a positive electrode empty foil area, herein the positive electrode coating area has macro-pores and micro-mesopores, the specific surface area of the macro-pores of the positive electrode coating area is 3.0˜7.0 m2/g, and the specific surface area of the micro-mesopores of the positive electrode coating is 2˜5 m2/g; and a negative electrode piece, comprising a negative electrode coating area and a negative electrode empty foil area, herein the negative electrode coating area has macro-pores and micro-mesopores, the specific surface area of the macro-pores of the positive electrode coating area is 0.8˜2.0 m2/g, and the specific surface area of the micro-mesopores of the positive electrode coating is 0.6˜1.7 m2/g.

Abstract: A computing system comprising a memory configured to store an artificial intelligence (AI) model and an image, and a computation engine executing one or more processors may be configured to perform the techniques for error-based explanations for AI behavior. The computation engine may execute the AI model to analyze the image to output a result. The AI model may, when analyzing the image to output the result, process, based on data indicative of the result, the image to assign an error score to each image feature extracted from the image, and obtain, based on the error scores, an error map. The AI model may next update, based on the error map and to obtain a first updated image, the image to visually indicate the error score assigned to each of the image features, and output one or more of the error scores, the error map, and the first updated image.