Abstract: First event data, indicative of a first activity on a computer network and second event data indicative of a second activity on the computer network, is received. A first machine learning anomaly detection model is applied to the first event data, by a real-time analysis engine operated by the threat indicator detection system in real time, to detect first anomaly data. A second machine learning anomaly detection model is applied to the first anomaly data and the second event data, by a batch analysis engine operated by the threat indicator detection system in a batch mode, to detect second anomaly data. A third anomaly is detected using an anomaly detection rule. The threat indictor system processes the first anomaly data, the second anomaly data, and the third anomaly data using a threat indicator model to identify a threat indicator associated with a potential security threat to the computer network.

Abstract: Systems, computer-implemented methods, and computer program products to facilitate entity steering of a running quantum program are provided. According to an embodiment, a system can comprise a processor that executes computer executable components stored in memory. The computer executable components can comprise a steering component that adjusts at least one parameter corresponding to a running quantum program to define at least one modified parameter. The computer executable components can further comprise an execution component that executes one or more shots of the running quantum program based on the at least one modified parameter.

Abstract: Techniques disclosed herein relate to learning and applying contextual patient similarities. Multiple template similarity functions (118) may be provided (602). Each template similarity function may compare a respective subset of features of a query entity feature vector with a corresponding subset of features of a candidate entity feature vector. A composite similarity function (120) may be provided (604) as a weighted combination of respective outputs of the template similarity functions. A plurality of labeled entity vectors may be provided (606) as context training data. An approximation function may be applied (608) to approximate a first context label for each respective labeled entity vector. A first context specific composite similarity function may be trained (610) based on the composite similarity function by learning first context weights for the template similarity functions using a first loss function based on output of application of the approximation function to the first context training data.

Abstract: The approaches presented herein may include loading a plurality of metadata instances for a plurality of data blocks stored in a database, each metadata instance of the plurality of metadata instances indicating a respective data block structure applied to aggregate a set of data messages. The approaches may include processing a first metadata instance corresponding to a first set of one or more data blocks of the plurality of data blocks and a second metadata instance corresponding to a second set of one or more data blocks of the plurality of data blocks to perform an anomaly storage check between the first set of one or more data blocks of the plurality of data blocks and the second set of one or more data blocks of the plurality of data blocks. The approaches may include generating an output based at least in part on the anomaly storage check.

Abstract: An example method comprises receiving historical sensor data of a first time period, the historical data including sensor data of a renewable energy asset, extracting features, performing a unsupervised anomaly detection technique on the historical sensor data to generate first labels associated with the historical sensor data, performing at least one dimensionality reduction technique to generate second labels, combining the first labels and the second labels to generate combined labels, generating one or more models based on supervised machine learning and the combined labels, receiving current sensor data of a second time period, the current sensor data including sensor data of the renewable energy asset, extracting features, applying the one or more models to the extracted features of the current sensor data to create a prediction of a future fault in the renewable energy asset, and generating a report including the prediction of the future fault in the energy asset.

Abstract: System, apparatus and method may permit users to collaboratively engage in inference on a computer and visualize structure of that inference, and provide a formal verification system for informal argumentation and inference. The system and method may generate and allow for modification of graphical structures that represent sequences of structured rational argumentation; and automatically monitor, compute and represent ratings or scores of nodes within the structure; indicate whether a node is supported by a chain of argumentation that has not been validly rebutted. The graphical structures may be displayed to bring into focus contentious and significant underlying points within an argument, and simulate the effects of alternative resolutions of these contentious points. The graphical displays may provide a transparent verification to other users of the state of what can be demonstrated and refuted, allow discovery of weak or missing points in a logical argument, and allow rational inference by users.

Abstract: A system and method for updating and correcting facts that receives proposed values for facts from users and determines a correctness score which is used to automatically accept or reject the proposed values.

Abstract: In exemplary aspects, a golden data structure can be used to validate the stability of machine learning (ML) models and weights. The golden data structure includes golden input data and corresponding golden output data. The golden output data represents the known correct results that should be output by a ML model when it is run with the golden input data as inputs. The golden data structure can be stored in a secure memory and retrieved for validation separately or together with the deployment of the ML model for a requested ML operation. If the golden data structure is used to validate the model and/or weights concurrently with the performance of the requested operation, the golden input data is combined with the input data for the requested operation and run through the model. Relevant outputs are compared with the golden output data to validate the stability of the model and weights.

Abstract: In one embodiment, a method includes receiving, by one or more processors of an information processing system, one or more results of a task. The method includes determining, by the one or more processors, an accuracy confidence score for each result of the task. The method includes determining, by the one or more processors, that the results satisfy an accuracy quality threshold set for the task based on the accuracy confidence score for each result of the task. The method includes providing, by the one or more processers, a determination for the task to a customer system in response to the determination that the results satisfy the accuracy quality threshold.

Abstract: Computer hardware and/or software that performs the following operations: (i) assessing a performance of a plurality of unsupervised machine learning pipelines against a plurality of data sets; (ii) associating the performance with meta-features corresponding to respective pipeline/data set combinations; (iii) training a supervised meta-learning model using the associated performance and meta-features as training data; and (iv) utilizing the trained model to identify one or more pipelines for processing an input data set.

Abstract: Methods and systems are disclosed for updating learned models. An embodiment comprises receiving a plurality of data sets representing sensed data from one or more devices and determining, using one or more local learned models, local parameters based on the received data sets. Another operation may comprise generating a combined data set by combining the plurality of data sets and, determining, using one or more local learned models, global parameters based on the combined data set. Another operation may comprise transmitting, to a remote system, the global parameters for determining updated global parameters using one or more global learned models based at least partially on the global parameters, and receiving, from the remote system, the updated global parameters. Another operation may comprise updating the one or more local learned models using both the local parameters and updated global parameters.

Abstract: Methods, apparatuses, and embodiments related to analyzing the content of digital images. A computer extracts multiple sets of visual features, which can be keypoints, based on an image of a selected object. Each of the multiple sets of visual features is extracted by a different visual feature extractor. The computer further extracts a visual word count vector based on the image of the selected object. An image query is executed based on the extracted visual features and the extracted visual word count vector to identify one or more candidate template objects of which the selected object may be an instance. When multiple candidate template objects are identified, a matching algorithm compares the selected object with the candidate template objects to determine a particular candidate template of which the selected object is an instance.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a neural network system used to control an agent interacting with an environment. One of the methods includes receiving a current observation; processing the current observation using a proposal neural network to generate a proposal output that defines a proposal probability distribution over a set of possible actions that can be performed by the agent to interact with the environment; sampling (i) one or more actions from the set of possible actions in accordance with the proposal probability distribution and (ii) one or more actions randomly from the set of possible actions; processing the current observation and each sampled action using a Q neural network to generate a Q value; and selecting an action using the Q values generated by the Q neural network.

Abstract: A device may receive network data associated with a network, and may calculate, based on the network data, key performance indicators (KPIs) for the network. The device may generate a first user interface that depicts one or more of the KPIs, and may receive a selection of a particular KPI from the one or more KPIs displayed by the first user interface. The device may parse a set of rules, utilized to calculate the particular KPI, to generate a parsed set of rules, and may analyze the parsed set of rules to identify particular metrics utilized to calculate the particular KPI. The device may generate a second user interface that depicts one or more timeline views of the particular metrics correlated with the particular KPI, and may provide the second user interface for display.

Abstract: A system may provide for the design and/or modification of network slices associated with a wireless network. The wireless network may include different slices that are associated different sets of service parameters. Slices may include radio access networks (“RANs”), core networks, or other types of networks, which may include respective sets of network functions (“NFs”), which may perform specific functions with respect to a given RAN and/or core network. Different slices, RANs, core networks, and/or NFs may be associated with particular policies and/or tags which may be specified by one or more users associated with a first access level. One or more users associated with a second access level may configure portions of the wireless network, and the policies and/or tags associated with particular slices, RANs, core networks, or NFs may be automatically implemented by an orchestration system that configures the wireless network based on the provided configuration information.

Abstract: In one embodiment, a device obtains page load information corresponding to a loaded web application. The device detects, based on the page load information, an anomalous feature of the loaded web application. The device identifies a type of the anomalous feature based on a number of resource anomalies within the loaded web application, wherein the type of the anomalous feature is selected from a group consisting of: a page anomaly; a resource anomaly; and a domain anomaly. The device performs one or more mitigation actions according to the type of the anomalous feature.

Abstract: System derives training change factors for services provided for training product user, priority assigned to training service ticket initiated by training product user, times of service ticket interactions associated with training service ticket, and/or age of training service ticket, and also for times of states of training service ticket. System uses training service ticket and training change factors to train change-based machine-learning model to predict change-based training probability that training product user escalated service for training service ticket. System derives change factors for services provided for product user, priority assigned to service ticket initiated by product user, times of service ticket interactions associated with service ticket, and/or age of service ticket, and also for times of states of training service ticket.

Abstract: Certain aspects of the present disclosure provide techniques for training decision trees representing users of a software application. An example method generally includes generating, from a transaction history data set for a plurality of users of a software application, a plurality of grouped data sets including transactions grouped by counterparty. A plurality of feature vectors are generated from the plurality of grouped data sets. Each feature vector generally corresponds to a user of the plurality of users and includes a plurality of features describing relationships between the user and a plurality of counterparties in a transaction history associated with the user. A decision tree is trained based on the plurality of feature vectors. The decision tree generally includes a plurality of paths terminating in a similar or different classification, and the plurality of paths distinguishes a user associated with the decision tree from other users of the software application.

Abstract: A computer system is provided that is programmed to select feature sets from a large number of features. Features for a set are selected based on metagradient information returned from a machine learning process that has been performed on an earlier selected feature set. The process can iterate until a selected feature set converges or otherwise meets or exceeds a given threshold.

Abstract: A machine learning model training method includes: training a machine learning model using features of samples in a training set, where a sample in the training set corresponds to an initial first weight and an initial second weight. In one iteration, the method includes: determining a first sample set comprising one or more samples whose corresponding target variables are incorrectly predicted; determining an overall predicted loss of the first sample set based on the predicted losses and corresponding first weights of samples in the first sample set; updating the first weights and second weights of the samples in the first sample set based on the overall predicted loss of the first sample set; and inputting the second weights, the features, and the target variables of the samples in the training set to the machine learning model, and initiating a next iteration of training the machine learning model.

Abstract: A meta-learning system includes an inner function computation module, adapted to compute output data from applied input data according to an inner model function, depending on model parameters; an error computation module, adapted to compute errors indicating mismatches between the computed output data and target values; a state update module, adapted to update the model parameters of the inner model function according to an updated state, updated based on a current state of the state update module, in response to an error received from the error computation module. The state update module is learned to adjust the model parameters of the inner model function, such that a following training of the inner model function with training data is improved.

Abstract: A computer implemented method to detect anomalous behavior of a software container having a software application executing therein, the method including receiving a sparse data representation of each of a: first set of container network traffic records; a first set of application traffic records; and a first set of container resource records, and training an hierarchical temporal memory (HTM) for each first set, wherein the container network traffic records correspond to network traffic communicated with the container, the application traffic records correspond to network traffic communicated with the software application, and the container resource records correspond to the use of computer resources by the container; receiving a sparse data representation of each of a: second set of container network traffic records; a second set of application traffic records; and a second set of container resource records; executing the trained HTMs based on each respective second set to determine a degree of recognition o

Abstract: A growth transform neural network system is disclosed that includes a computing device with at least one processor and a memory storing a plurality of modules, including a growth transform neural network module, a growth transform module, and a network convergence module. The growth transform neural network module defines a plurality of mirrored neuron pairs that include a plurality of first components and a plurality of second components. Each first and second component is connected by a normalization link. The first components are interconnected according to an interconnection matrix, and the second components are interconnected according to the interconnection matrix. The growth transform module updates each first component of each mirrored neuron pair according to a growth transform neuron model. The network convergence module converges the plurality of mirrored neuron pairs to a steady state condition by solving a system objective function subject to at least one normalization constraint.

Abstract: Techniques for machine learning (ML) model training and deployment using model building blocks via graphical user interfaces (GUIs) are described. Users can use a GUI provided by an electronic device to select and configure ML aspects for one or more ML models to be trained using identified training data. The electronic device can send a request to cause a model construction service to train one or more ML models based on the user configuration, return results of the training to the user within the GUI, and deploy one or more of the ML models.

Abstract: An anomaly detection system is disclosed capable of reporting anomalous processes or hosts in a computer network using machine learning models trained using unsupervised training techniques. In embodiments, the system assigns observed processes to a set of process categories based on the file system path of the program executed by the process. The system extracts a feature vector for each process or host from the observation records and applies the machine learning models to the feature vectors to determine an outlier metric each process or host. The processes or hosts with the highest outlier metrics are reported as detected anomalies to be further examined by security analysts. In embodiments, the machine learnings models may be periodically retrained based on new observation records using unsupervised machine learning techniques. Accordingly, the system allows the models to learn from newly observed data without requiring the new data to be manually labeled by humans.

Abstract: Disclosed herein are embodiments of systems, methods, and products comprises an analytic server, which evaluates user data for premium financing status and dynamically renders graphical user interfaces. The server trains an artificial intelligence model based on historical user data. The artificial intelligence model comprises one or more data points with each data point representing one of a plurality of attributes and applies a logistic regression algorithm to identify a weight factor for each attribute. The server uses a dynamic algorithm to generate a score by combining the plurality of attributes based on the weight factors. The server receives responses regarding the scores that indicate the premium financing status of each case. The server retrains the artificial intelligence model to identify new weight factors based on negative responses data. The server automatically displays new scores calculated based on the new weight factors.

Abstract: Systems and methods are disclosed to implement a chart recommendation system that recommends charts to users during a chart building process. In embodiments, when a new chart is being created, specified features of the chart are provided to a machine learned model such as a self-organizing map. The model will determine a previous chart that is the most similar to the new chart and recommend the previous chart to the user for recreation. In embodiments, newly created charts are added to a library and used to update the model. Charts that are highly popular or authored by expert users may be weighed more heavily during model updates, so that the model will be more influenced by these charts. Advantageously, the disclosed system allows novice users to easily find similar charts created by other users. Additionally, the disclosed system is able to automatically group similar charts without using human-defined classification rules.

Abstract: Techniques are described for identifying resource bottlenecks in decomposing monolithic software applications as part of software modernization processes. An application modernization system constructs a graph model of a software application based on an analysis of application artifacts associated with the software application. The graph model includes nodes representing independent application components, and further includes edges representing identified dependency relationships among the application components. An application modernization system further generates application profile metrics associated with the identified dependencies, and weights derived from the metrics are applied to the nodes and/or the edges of the graph model to generate a weighted graph model that identifies the resource bottlenecks among the application components and the identified dependency relationships. The weighted graph model is transmitted to a computing device for display to a user.

Abstract: Techniques are described for providing live first aid response guidance using a machine learning based cognitive aid planner. In one embodiment, a computer-implemented method is provided that comprises classifying, by a system operatively coupled to a processor, a type of an injury endured by a patient. The method further comprises, employing, by the system, one or more machine learning models to estimate a risk level associated with the injury based on the type of the injury and a current context of the patient.

Abstract: Aspects of the present disclosure involve a system comprising a computer-readable storage medium storing a program and a method for performing operations comprising: receiving a plurality of records associated with a first geographical area; identifying a plurality of corrections to a first attribute in the first geographical area in the plurality of records for a particular time period; based on identifying the plurality of corrections to the first attribute, computing a first metric representing a quantity of the plurality of corrections to the first attribute per effort during the particular time period; accumulating a first value representing a total number of errors across a plurality of time periods up to and including the particular time period based on the identified plurality of corrections; and generating a first model that predicts accuracy of the first attribute in the first geographical area based on the metric and the accumulated first value.

Abstract: The present disclosure provides a method for managing a parking lot in a smart city based on an Internet of Things, which is executed by a management platform. The method comprises obtaining a user position of a user platform based on a service platform, determining a candidate parking lot that meets a preset condition; determining time when a vehicle to be parked arrives at the candidate parking lot based on the user position; determining free parking space information when the vehicle to be parked arrives at the candidate parking lot; determining recommendation information based on the free parking space information; and sending the recommendation information to the user platform based on the service platform.

Abstract: A mechanism is provided in a data processing system for preventing data leakage in automated machine learning. The mechanism receives a data set comprising a label for a target variable for a classifier machine learning model and a set of features. For each given feature in the set of features, the mechanism trains a subprime classifier model using the given feature as a target variable and remaining features as independent input features, tests the subprime classifier model, and records results of the subprime classifier model. The mechanism performs statistical analysis on the recorded results to identify an outlier result corresponding to an outlier subprime classifier model.

Abstract: A model validation system is described that is configured to automatically validate model artifacts corresponding to models. For a model artifact being validated, the model validation system is configured to dynamically determine the validation checks to be performed for the model artifact, where the validation checks include various validation checks to be performed at the model artifact level and also for individual components included in the model artifact. The checks to be performed are dynamically determined based upon the attributes of the model artifact and of the components within the model artifact. The system is configured to generate a validation report that comprises information regarding the checks performed and the results generated from performing the various validation checks. The validation report may also include information suggesting actions for passing checks that result in a failed check, or for improving the scores of certain validation checks.

Abstract: The disclosure provides a method for detecting and defending a Distributed Denial of Service attack in an SDN environment. The method includes: building data messages acquired as feature messages by a proxy module; sending the feature messages to a pre-built detection model to obtain a detection result; making a decision instruction based on the detection result; and performing control operations by the proxy module based on the decision instruction.

Abstract: An embodiment may involve obtaining a set of pre-defined features and a new document; extracting a subset of the pre-defined features from within new document; applying a natural language model to the new document, wherein the natural language model was pre-trained using scientific or medical literature and fine-tuned using a corpus of documents; applying a feature-based model to the subset of the pre-defined features extracted from the new document, wherein the feature-based model was trained with the pre-defined features and the respective labels of the documents; and applying an aggregation model to the classifications of the new document produced by the natural language model and the feature-based model, wherein the aggregation model was trained with prior classifications produced by the natural language model and the feature-based model so that the aggregation model produces a further classification of the new document representing its relevance to pharmacovigilance.

Abstract: A system includes at least one waveguide with optical interaction portions that represent a sequence of reference symbols based on the order of the optical interaction portions in the at least one waveguide. At least one light source sends photons of one or more predetermined wavelengths into the at least one waveguide representing a string of query symbols. A detector detects photons received from the at least one waveguide that result from optical interactions between photons sent into the at least one waveguide and one or more corresponding optical interaction portions with an interaction between photons and an optical interaction portion indicating a match between a query symbol and a reference symbol. An analyzer determines one or more respective relative locations of the one or more corresponding optical interaction portions indicating one or more relative locations of the string of query symbols in the reference sequence.

Abstract: System and method for predicting and scoring a data model are provided. The system includes a memory configured to receive a plurality of data sets. The system also includes a processing subsystem operatively coupled to the memory and configured to select one or more variables based on a plurality of parameters, to apply feature engineering and transformation on one or more variables to extract a plurality of features from the plurality of data sets, to build new data model based on the plurality of features, to evaluate a classification technique to select a right machine learning model based on a plurality of elements, to predict a newly built data model based on an evaluated classification technique and score the predicted data model. The system further includes a display model operatively coupled to the processing subsystem and configured to present the predicted and scored data model in one or more forms.

Abstract: Method and system for classifying and labeling images, which can perform segmentation based on features of each part of images, classify and match the image and the segmented image based on a classification model built by the machine learning method. Meanwhile, each image is assigned with labels and text descriptions. The system also includes a string module assigning the image with a plurality of matching labels and text descriptions that are the most relevant in recent times. Furthermore, the classification model is trained by machine learning method such as an unsupervised learning, a self-supervised learning, or a heuristic algorithms. In addition, a character recognition module is provided to extract characters in the image for comprehensive learning and calculations to facilitate classification and labeling of the image.

Abstract: Devices and techniques are generally described for content classification. In some examples, first item data representing a first item may be received. The first item data may include a plurality of prediction scores output by a machine learning model. Each prediction score of the plurality of prediction scores may be associated with a respective label of a plurality of labels. In some examples, a set of one or more labels among the plurality of labels may be predicted. The set of labels may be predicted as being applicable to the first item for classification of the first item. A determination may be made that the set of one or more labels represents a complete set of labels applicable to the first item. In some examples, the first item may be classified based on the set of one or more labels.

Abstract: Systems and methods for a cognitive system to interact with a user are provide. Aspects include receiving a cognitive system profile and observational data associated with the user. Environmental data associated with the user is received and features are extracted from the observations data and the environmental data. The features are stored in the user profile and analyzed to determine a situational context for each of the features based on the cognitive system profile and the user profile. Trigger events are identified based on the situational context for each of the features. One or more proposed actions are determined based at least in part on the one or more trigger events. At least one action is initiated from the one or more proposed actions and are stored in the user profile along with the one or more trigger events and the one or more features.

Abstract: Disclosed is a technique that can be performed by an electronic device. The electronic device can generate time-stamped events, extract training data from the time-stamped events, and send the training data over a network to a remote computer. The electronic device can receive model data generated by the remote computer from the training data by use of a machine learning process, update a local model of the electronic device based on the received model data, and generate an output by processing locally sourced data of the electronic device with the updated local model.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for reinforcement learning. The embodiments described herein apply meta-learning (and in particular, meta-gradient reinforcement learning) to learn an optimum return function G so that the training of the system is improved. This provides a more effective and efficient means of training a reinforcement learning system as the system is able to converge on an optimum set of one or more policy parameters ? more quickly by training the return function G as it goes. In particular, the return function G is made dependent on the one or more policy parameters ? and a meta-objective function J? is used that is differentiated with respect to the one or more return parameters ? to improve the training of the return function G.

Abstract: The present disclosure provides systems and methods for training probabilistic object motion prediction models using non-differentiable representations of prior knowledge. As one example, object motion prediction models can be used by autonomous vehicles to probabilistically predict the future location(s) of observed objects (e.g., other vehicles, bicyclists, pedestrians, etc.). For example, such models can output a probability distribution that provides a distribution of probabilities for the future location(s) of each object at one or more future times. Aspects of the present disclosure enable these models to be trained using non-differentiable prior knowledge about motion of objects within the autonomous vehicle's environment such as, for example, prior knowledge about lane or road geometry or topology and/or traffic information such as current traffic control states (e.g., traffic light status).

Abstract: Various techniques are described for automatically suggesting variation parameters used to generate a tailored synthetic dataset to train a particular machine learning model. A seeding taxonomy associates a plurality of machine learning scenarios with corresponding subsets of variation parameters. A selected machine learning scenario is used to retrieve a corresponding subset of variation parameters associated with the selected machine learning scenario by the seeding taxonomy. The seeding taxonomy may be adaptable using a feedback loop that tracks selected variation parameters and updates the seeding taxonomy. The suggested variation parameters are presented as suggestions to assist users to identify and select relevant variation parameters faster and more efficiently. Further embodiments relate to pre-packaging synthetic datasets for common or anticipated machine learning scenarios.

Abstract: A computer system obtains information describing a geographical area segmented into multiple first clusters serviced by a telecommunications network. Multiple test locations are identified within the first clusters. Each test location is located within a grid of the geographical area. Each first cluster is recursively segmented into multiple second clusters until a difference between a number of test locations within each second cluster and a target number of test locations is less than a threshold number of test locations. A route is generated connecting test locations within each second cluster, using a routing application programming interface for performing drive testing of the telecommunications network. The computer system sends the route to one or more computer devices for performing the drive testing at the test locations in a sequence corresponding to the route.

Abstract: The present disclosure relates to a computer-implemented method for automatically determining pin placement associated with an electronic design. Embodiments may include receiving, using at least one processor, at least one layout associated with the electronic design and separating the at least one layout into one or more grids. Embodiments may also include extracting one or more connectivity features from the one or more grids, wherein the one or more connectivity features include instance-pin and pin information. Embodiments may also include training a machine learning model, based upon, at least in part, the one or more connectivity features and receiving the machine learning model and a test layout at a predictor engine. Embodiments may further include providing a user with a pin placement recommendation based upon, at least in part, the machine learning model and the test layout.

Abstract: A sound data processing method includes acquiring sound data of a target by input. The sound data processing method further includes: generating similar sound data that becomes a similar sound similar to the sound data of the target, based on the sound data of the target; and performing machine learning by using the acquired sound data of the target and the generated similar sound data as learning sound data, and generating a learning model for performing classification determination related to the sound data of the target.

Abstract: Disclosed are a method and an apparatus for identifying animal species by using audiovisual information. A method for identifying animal species, according to one embodiment of the present invention, may include: a step of receiving an input signal for an object to be identified; a step of processing image information and acoustic information based on the input signal, wherein a processing result of the image information and a processing result of the acoustic information are represented by class-specific scores; a step of determining whether the image information processing result and the acoustic information processing result corresponding to the input signal exist; and a final result derivation step of fusing the image information processing result and the acoustic information processing result according to the determination result and classifying the object to be identified as a certain animal species by using the fused processing result.

Abstract: One example of a system comprises using a processor for identifying a model to be validated that is stored in a repository; automatically computing and recording one or more model metrics for the model to be validated in a tamper-proof manner; comparing the computed tamper-proof metrics with one or more encoded rules and policies to determine if the model to be validated complies with the one or more encoded rules and policies; and outputting a notification to a device indicating a validation status of the model to be validated based on the comparison of the computed tamper-proof metrics with the one or more encoded rules and policies.

Abstract: A method performed by one or more processors that preserves a machine learning model comprises accessing model parameters associated with a machine learning model. The model parameters are determined responsive to training the machine learning model. The method comprises generating a plurality of model parameter sets, where each of the plurality of model parameter sets comprises a separate portion of the set of model parameters. The method comprises determining one or more parity sets comprising values calculated from the plurality of model parameter sets. The method comprises distributing the plurality of model parameter sets and the one or more parity sets among a plurality of computing devices, where each of the plurality of computing devices stores a model parameter set of the plurality of model parameter sets or a parity set of the one or more parity sets. The method comprises accessing, from the plurality of computing devices, a number of sets comprising model parameter sets and at least one parity set.