Abstract: A method for selective performing a validation in one or more technical operations by a cloud based validation framework is provided. The method includes identifying a job task to be performed, identifying technical operations associated with the job task; receiving user configurable input data for performing validation on at least one technical operation associated with the job task; and submitting the user configurable input data into the cloud network based validation framework. The method further includes determining, by the cloud network based validation framework, at least one technical operation specified for performing a validation function among the technical operations; generating a dictionary file for the at least one technical operation specified for performing the validation function; and performing the validation function specified by the dictionary file.

Abstract: An end-to-end automatic speech recognition (ASR) system can be constructed by fusing a first ASR model with a transformer. The input of the transformer is a learned layer generated by the first ASR model. The fused ASR model and transformer can be treated as a single end-to-end model and trained as a single model. In some embodiments, the end-to-end speech recognition system can be trained using a teacher-student training technique by selectively truncating portions of the first ASR model and/or the transformer components and selectively freezing various layers during the training passes.

Abstract: A method of distributed data detection may include: obtaining, by an end-unit, a plurality of datasets; streaming, by the end-unit to a remote computing device, at least a part of the obtained datasets; determining, by the end-unit, end-unit related information and sending the end-unit related information to the remote computing device; receiving, by the remote computing device, a desired data portion related information; selecting, by the remote computing device, based on at least a part of the obtained datasets and at least one of: the end-unit related information and the desired data portion related information, an operation to be used by the end-unit to detect the desired data portion in at least a part of the obtained datasets; and detecting, by the end-unit, the desired data portion in at least a part of the obtained datasets using the selected operation.

Abstract: Techniques for synchronizing the authentication state of a user across data centers using event messages are provided herein. In one example, a system can determine a first change in the authentication state, update a first data structure representing the authentication state in a first data center based on the change, and store a first event message indicating the first change in a message queue. The system can then transmit the first event message to a second data center. Based on the event message, the second data center can update a second data structure also representing the authentication state of the user. In some examples, the second data center can also transmit a second event message to the first data center indicating a second change to the authentication state of the user. The first data center can update first data structure based on the second event message.

Abstract: Described herein is an apparatus and method for automating interactions. In some embodiments, apparatus may gather system data, determine event activation data as a function of system data, execute an event by communicating event activation data to an external device, and update system data based on execution of an event.

Abstract: The present disclosure relates generally to methods, systems and computer program products for classifying and identifying input data using neural networks and displaying results (e.g., images of vehicles, vehicle artifacts and geographical locations dating from the 1880s to present day and beyond). The results may be displayed on displays or in virtual environments such as on virtual reality, augmented reality and/or mixed-reality devices.

Abstract: A method including providing a graph database representation, wherein the graph database representation represents a plurality of nodes in a graph which are interconnected by respective edges, wherein each node of the plurality of the nodes represents a data sample and is assigned to at least one node feature, wherein each edge of the plurality of the edges represents a relationship between the data samples, transforming the graph database representation into a data matrix using a first machine learning algorithm suitable for graph data and an architecture as first layers of a joint machine learning architecture, determining at least one node of the plurality of nodes based on the transformed data matrix using a second machine learning algorithm and a second architecture as second layers of a joint machine learning architecture, and providing the at least one determined node. Further, a computing unit and a computer program product is provided.

Abstract: A method includes: performing a plurality of times of first clustering processing on a plurality of training images based on a plurality of target features to obtain a plurality of first clustering results, where each of the plurality of first clustering results corresponds to one silhouette coefficient, and the silhouette coefficient indicates cluster quality; determining a first target clustering result based on the silhouette coefficients, where the first target clustering result includes M clustering categories; and performing second clustering processing on the plurality of training images based on the plurality of target features to obtain a second clustering result, where a quantity of clustering categories included in the second clustering result is M.

Abstract: Candidate material for polymerization can be received. One or more desired features in the candidate material can be identified. A machine learning model can be trained to generate a new material having one or more of the desired features. Permissively, the candidate material can be determined from running a machine learning classification model that ranks a plurality of material as candidates. Permissively, the generated new material can be input to the machine learning classification model, for the machine learning classification model to include in ranking the plurality of material as candidates.

Abstract: A bleed air supply system has a sensor monitoring the system, an operating condition monitor detecting an operating condition value of the aircraft (other than the bleed air supply system), and independent monitoring modules evaluating a part of the bleed air supply system. The monitoring modules each have an individual monitoring function and individual activation and deactivation parameters based on sensor data and the operating condition value. The method includes: detecting the condition of the bleed air supply system via sensor data and the operating condition value; activating a monitoring module, which has activation parameters met by the sensor data and the operating condition value; monitoring the condition of the bleed air supply system by the activated monitoring module by a monitoring function of the activated monitoring module; and deactivating the activated monitoring module, deactivation parameters of which are met by the sensor data and the operating condition value.

Abstract: Disclosed are a system, apparatus and techniques for evaluating a dataset to confirm that the data in the dataset satisfies a data quality metric. A machine learning engine or the like may evaluate text strings within the dataset may be of arbitrary length and encoded according to an encoding standard. Data vectors of a preset length may be generated from the evaluated text strings using various techniques. Each data vector may be representative of the content of the text string and a category may be assigned to the respective data vector. The category assigned to each data vectors may be evaluated with respect to other data vectors in the dataset to determine compliance with a quality metric. In the case that a number of data vectors fail to meet a predetermined quality metric, an alert may be generated to mitigate any system errors that may result from unsatisfactory data quality.

Abstract: A method of managing a desired state of a software-defined data center (SDDC) includes the steps of: receiving an original desired state document that includes configurations and associated criteria for applying the configurations; evaluating a first criteria to determine that a first configuration associated with the first criteria is applicable to components of the SDDC; evaluating a second criteria to determine that a second configuration associated with the second criteria is not applicable to any components of the SDDC; creating an updated desired state of the SDDC, as a result of the evaluating of the first and second criteria, the updated desired state including the first configuration and excluding the second configuration; and applying the updated desired state to the SDDC.

Abstract: A photonic neural network device may include a planar waveguide; a layer having a changeable refractive index adjacent to the planar waveguide; and a plurality of electrodes. Each electrode may be electrically coupled to the layer having the changeable refractive index at a corresponding location of the layer having the changeable refractive index. Each electrode may be configured to apply a corresponding, configurable voltage to the corresponding location to affect a refractive index of the corresponding location of the layer having the changeable refractive index to induce an amplitude modulation or a phase modulation of a light waveform propagating through the photonic neural network device to configure a corresponding neuron of the photonic neural network device in order to perform a computation.

Abstract: Provided are a training method for an air quality prediction model, a prediction method and apparatus, a device, a program, and a medium. The method includes the steps described below. A target monitoring range is divided into a plurality of regions; the air quality prediction model is pre-trained by adopting a pre-training sample and a pre-training objective function, where the pre-training sample includes measurement values; and the pre-trained air quality prediction model is trained by adopting a formal training sample and a formal training objective function, where the formal training sample includes the measurement values. The air quality prediction model is configured to predict air quality of the plurality of regions according to spatial information, historical information and environmental information.

Abstract: In an auto-classification system, example documents whose content exemplifies a content category or classification can be imported into a classification model. The classification model is tested to assess accuracy. Based on the testing, metrics or other information can be provided as feedback to a user. The user can iteratively refine the classification model and keep re-running the classifications to view how each change to the classification model improves accuracy. If no user refinement is desired, the auto-classification system classifies documents utilizing the classification model. This technology enhances the overall transparency and defensibility of the auto-classification process.

Abstract: A set of distance measurable encrypted feature vectors can be derived from any biometric data and/or physical or logical user behavioral data, and then using an associated deep neural network (“DNN”) on the output (i.e., biometric feature vector and/or behavioral feature vectors, etc.) an authentication system can determine matches or execute searches on encrypted data. Behavioral or biometric encrypted feature vectors can be stored and/or used in conjunction with respective classifications, or in subsequent comparisons without fear of compromising the original data. In various embodiments, the original behavioral and/or biometric data is discarded responsive to generating the encrypted vectors. In another embodiment, distance measurable or homomorphic encryption enables computations and comparisons on cypher-text without decryption of the encrypted feature vectors. Security of such privacy enabled embeddings can be increased by implementing an assurance factor (e.g.

Abstract: An embodiment includes identifying, from a training dataset for training a model, a first unlabeled datapoint to present for labelling according to a first query strategy. The embodiment also includes issuing a query requesting a label for the first unlabeled datapoint. The embodiment also includes receiving a labeled datapoint in response to the query, the labeled datapoint comprising the first unlabeled datapoint as labeled by an oracle. The embodiment also includes generating a causal network based on labeled datapoints from the training dataset. The embodiment also includes receiving an instruction to modify the causal network. The embodiment also includes replacing the first query strategy with a second query strategy based on the instruction to modify the causal network. The embodiment also includes identifying, from the training dataset, a second unlabeled datapoint to present for labelling according to the second query strategy.

Abstract: The disclosed technology receives a control input identifying a sampling criterion for classifying a data store storing a set of data objects in a computing environment as corresponding to a target data type and deploys one or more scanners configured to select a representative subset of data objects, from the set of data objects, based on the sampling criterion. A scanner result generated by the one or more scanners is received that represents detected instances, in the representative subset of data objects, of one or more pre-defined data patterns of the target data type. A classification result is generated based on a comparison of the number of detected instances of the one or more pre-defined data patterns to a threshold. The classification result represents a classification of the data store as having correspondence to the target data type. A computing action is performed based on the classification result.

Abstract: A method may include receiving, via a computing system, a request to generate one or more visualizations based on one or more datasets provided by an industrial device within an industrial system. The computing system may be communicatively coupled to the industrial device via a data backplane. The method may also involve accessing a visualization platform system in response to receiving the request, sending design data associated with the one or more visualizations to the visualization platform system, and receiving a visualization project file for execution by the computing system from the visualization platform system. The visualization project file may be generated based on the design data. The method may also include generating the one or more visualizations based on the visualization project file and presenting the one or more visualizations via an electronic display device.

Abstract: Systems and methods may ethically evaluate intelligent systems operating in a real-world environment. The systems and methods may generate a clone of the intelligent system, and test the clone in a simulation environment. If the clone passes the testing, the systems and methods may permit the intelligent system to continue operating in the real-world environment. If the clone fails the testing, the systems and methods may override the intelligent system, such as disabling the intelligent system and assuming control in the real-world environment. The systems and methods may be implemented at a hardware level of a data processing device to prevent interference with the systems and methods by the intelligent system.

Abstract: Systems and methods for natural language processing are described. Embodiments of the present disclosure receive text including an event trigger word indicating an occurrence of an event; classify the event trigger word to obtain an event type using a few-shot classification network, wherein the few-shot classification network is trained by storing first labeled samples during a first training iteration and using the first labeled samples for computing a loss function during a second training iteration that includes a support set with second labeled samples having a same ground-truth label as the first labeled samples; and transmit event detection information including the event trigger word and the event type.

Abstract: Methods and systems are presented for assisting a user to identify and evaluate features for use in a machine learning model configured to perform a task. Based on graph data associated with a graph data structure, a user interface is provided on a device. Based on user inputs received via the user interface, a feature candidate for the machine learning model is determined. The feature candidate is associated with a particular way of traversing the graph data structure to obtain attribute values associated with one or more vertices and/or one or more edges in the graph data structure. Based on the attribute values, a value corresponding to the feature candidate can be calculated. The value can be used to evaluate the effectiveness of the feature candidate in performing the task. The feature candidate can then be incorporated into the machine learning model as one of the input features.

Abstract: In an implementation, a computer-implemented method, includes collecting, as collected integration flows (iFlows), published iFlows. Descriptions of the collected iFlows are extracted as extracted descriptions and the extracted descriptions are parsed. A list of one or more interchangeable operators is created. The collected iFlows are iterated through. Automated performance recommendations for a new iFlow are provided.

Abstract: Systems and methods for aggregating data. The system is configured to receive metadata from an interactive graphical user interface (GUI) of a user device, aggregate field values from the data stored on one or more databases based on the received metadata and generate filter instructions based on the received metadata. The system is further configured to transmit the aggregated field values and the filter instructions to the user device, receive a user-customized filter set and subscription request for a synthetic symbol associated with the user-customized filter set from the user device, and create the synthetic symbol responsive to the subscription request. Moreover, the system aggregates one or more data values from the data stored on the databases associated with the created synthetic symbol and generates instructions to display the data values on the interactive GUI in accordance with the user-customized filter set associated with the created synthetic symbol.

Abstract: A processor-implemented neural network method includes: determining a reference sample among sequential input samples to be processed by a neural network, the neural network comprising an input layer, one or more hidden layers, and an output layer; performing an inference process of obtaining an output activation of the output layer based on operations in the hidden layers corresponding to the reference sample input to the input layer; determining layer contraction parameters for determining an affine transformation relationship between the input layer and the output layer, for approximation of the inference process; and performing inference on one or more other sequential input samples among the sequential input samples using affine transformation based on the layer contraction parameters determined with respect to the reference sample.

Abstract: A method in a premises device is provided, where the premises device is configured with a plurality of analytics models and located in a premises environment. The method includes generating a dataset comprising surveillance data associated with a premises environment in which the premises device operates, determining metadata associated with the dataset, determining, for each analytics model of the plurality of analytics models, a corresponding predicted performance score based at least in part on the metadata, selecting a highest-scoring analytics model of the plurality of analytics models based at least in part on the corresponding predicted performance score, and analyzing the dataset using the highest-scoring analytics model to generate a prediction.

Abstract: [Problem] The estimation accuracy of a class of an object is to be effectively improved. [Means of Solution] Provided is an information processing device including an estimation unit that estimates, based on an input image, a class of an object that is present in a real environment corresponding to an imaging range of the input image, wherein the object includes an acoustically useful object having an acoustic feature useful for class estimation, and the estimation unit estimates a class of the acoustically useful object based on acoustic data collected from around the acoustically useful object.

Abstract: Enhanced methods for improving the performance of classifiers are described. A ground-truth labeled dataset is accessed. A classifier predicts a predicted label for datapoints of the dataset. A confusion matrix for the dataset and classifier is generated. A credibility interval is determined for a performance metric for each label. A first labels with a sufficiently large credibility interval is identified. A second label is identified, where the classifier is likely to confuse, in its predictions, the first label with the second label. The identification of the second label is based on instances of incorrect label predictions of the classifier for the first and/or the second labels. The classifier is updated based on a new third label that includes an aggregation of the first label and the second label. The updated classifier model predicts the third label for any datapoint that the classifier previously predicted the first or second labels.

Abstract: A method and system is provided for identifying patters in datasets by identifying delimited regions of feature-space in which patterns occur. The delimited regions are then combined into an ensemble able to make predictions based on the identified regions of feature-space. The method may be used for classification, for regression, for auto-encoding, for simulation, and for other applications of pattern detection.

Abstract: The disclosed system obtains records from a database, determines weights associated with the records, and obtains a first and a second force acting on a record among the records. The system defines a projection space based on the records, represents the record in the projection space by projecting the record into the projection space to obtain a projected record, and represents the first force and the second force in the projection space. The system repeatedly applies the first force and the second force to the projected record, thereby changing a position of the projected record in the projection space, until an equilibrium between the first force and the second force is reached. The system determines how closely the value associated with the record satisfies the first value associated with the criterion based on a distance between the projected record in the projection space at equilibrium and the first force.

Abstract: A computing system can include one or more non-transitory computer-readable media that collectively store a neural network including one or more layers with relaxed spatial invariance. Each of the one or more layers can be configured to receive a respective layer input. Each of the one or more layers can be configured to convolve a plurality of different kernels against the respective layer input to generate a plurality of intermediate outputs, each of the plurality of intermediate outputs having a plurality of portions. Each of the one or more layers can be configured to apply, for each of the plurality of intermediate outputs, a respective plurality of weights respectively associated with the plurality of portions to generate a respective weighted output. Each of the one or more layers can be configured to generate a respective layer output based on the weighted outputs.

Abstract: In some implementations, a device may identify an account associated with a user, and the account may be managed by an entity. The device may determine, using at least one machine learning model, a classification of the user that indicates a level of quality of a relationship between the user and the entity. The device may determine, based on the classification determined using the at least one machine learning model, one or more adjustments that are to be applied to one or more charges assessed to the account by the entity. The device may apply the one or more adjustments to the one or more charges.

Abstract: A firewall monitors network activity and stores information about that network activity in a network activity log. The network activity is analyzed to identify a potential threat. The potential threat is further analyzed to identify other potential threats that are related to the potential threat, and are likely to pose a future risk to a protected network. A block list is updated to include the potential threat and the other potential threats to protect the protected network from the potential threat and the other potential threats.

Abstract: An inference system trains and performs inference using a sparse neural network. The sparse neural network may include one or more layers, and each layer may be associated with a set of sparse weights that represent sparse connections between nodes of a layer and nodes of a previous layer. A layer output may be generated by applying the set of sparse weights associated with the layer to the layer output of a previous layer. Moreover, the one or more layers of the sparse neural network may generate sparse layer outputs. By using sparse representations of weights and layer outputs, robustness and stability of the neural network can be significantly improved, while maintaining competitive accuracy.

Abstract: An apparatus for generating a generalized linear model structure definition by generating a gradient boosted tree model and separating each decision tree into a plurality of indicator variables upon which a dependent variable of the generalized linear model depends.

Abstract: An exemplary discovery platform includes machine-learning techniques for using medical imaging data to study a phenotype of interest, such as complex diseases with weak or unknown genetic drivers. An exemplary method of identifying a patient subgroup of interest, comprises inputting a plurality of medical images obtained from a group of clinical subjects into a trained unsupervised machine-learning model to obtain a plurality of embeddings in a latent space, clustering the plurality of embeddings to generate one or more clusters of embeddings, identifying one or more patient subgroups corresponding to the one or more clusters of embeddings, and associating each patient subgroup of the one or more patient subgroups with a covariant to identify the patient subgroup of interest.

Abstract: Techniques and systems for a security service system configured with a sensor component including a machine learning (ML) malware classifier to perform behavioral detection on host devices. The security service system may deploy a sensor component to monitor behavioral events on a host device. The sensor component may generate events data corresponding to monitored operations targeted by malware. The system may map individual events from events data onto a behavioral activity pattern and generate process trees. The system may extract behavioral artifacts to build a feature vector used for malware classification and generate a machine learning (ML) malware classifier. The sensor component may use the ML malware classifier to perform asynchronous behavioral detection on a host device and process system events for malware detection.

Abstract: A method and system for vetting items being shipped across national boundaries using a new technology enabling an automated system is provided. The automated system screens items for shipping through customs and validating the items for shipment according to customs rules and regulations. The system identifies and applies the appropriate rules for customs and other responsible agencies pertaining to the eligibility of any item being imported into a particular country. The present invention utilizes an unconventional combination of image recognition technology, machine learning algorithms, and rule engine algorithms to categorize, identify, and apply the appropriate rules to each and every item being considered for importation to another country.

Abstract: A speech recognition platform configured to receive an audio signal that includes speech from a user and perform automatic speech recognition (ASR) on the audio signal to identify ASR results. The platform may identify: (i) a domain of a voice command within the speech based on the ASR results and based on context information associated with the speech or the user, and (ii) an intent of the voice command. In response to identifying the intent, the platform may perform a corresponding action, such as streaming audio to the device, setting a reminder for the user, purchasing an item on behalf of the user, making a reservation for the user or launching an application for the user. The speech recognition platform, in combination with the device, may therefore facilitate efficient interactions between the user and a voice-controlled device.

Abstract: Methods, apparatus, systems and articles of manufacture are disclosed that optimize workflows. An example apparatus includes an intent determiner to determine an objective of a user input, the objective indicating a task to be executed in an infrastructure, a configuration composer to compose a plurality of workflows based on the determined objective, a model executor to execute a machine learning model to create a confidence score relating to the plurality of workflows, and a workflow selector to select at least one of the plurality of workflows for execution in the infrastructure, the selection of the at least one of the plurality of workflows based on the confidence score.

Abstract: An imaging device having a function of processing an image is provided. The imaging device has an additional function such as image processing, can hold analog data obtained by an image capturing operation in a pixel, and can extract data obtained by multiplying the analog data by a predetermined weight coefficient. Difference data between adjacent light-receiving devices can be obtained in a pixel, and data on luminance gradient can be obtained. When the data is taken in a neural network or the like, inference of distance data or the like can be performed. Since enormous volume of image data in the state of analog data can be held in pixels, processing can be performed efficiently.

Abstract: The present disclosure is directed toward systems, methods, and non-transitory computer readable media for generating and searching a hybrid search index. In some embodiments, the disclosed systems generate a hybrid search index that comprises one or more content items stored at a content management system or at external network locations linked to the content management system via software connectors along with world state data associated with the one or more content items. The disclosed systems can generate a search result from the hybrid search index in response to receiving a search query of the hybrid search index. In some cases, the disclosed systems can rank one or more content items included in the search result based on observation layer data of the one or more content items.

Abstract: A system can receive a first set of data. The first set of data can include information indicating a first set of user sessions and for each of the first set of user sessions having an associated summary and a corresponding agent indicated intent. The system can also, based on the first set of data, determine a set of utterances and for each of the set of utterances a corresponding set of intents. Additionally, the system can receive a second set of data. The second set of data including information indicating a second set of user sessions and for each of the second set of user sessions having an associated determined utterance and corresponding interaction of a user. Moreover, the system can validate a corresponding intent of one or more utterances of the set of utterances, based on the second set of data.

Abstract: A method and system of training an interpretable deep learning model includes receiving an input set of data, which may be complex. The input set of data is provided to deep learning model for feature extraction. In an exemplary embodiment, the deep learning model generates a disentangled latent space of features from the feature extraction. The features may comprise semantically meaningful data which is then provided to a low-complexity learning model. The low-complexity learning model generates output based on a specified task (for example, classification or regression). Being a low-complexity learning model provides confidence that the data output from the deep learning model is inherently interpretable.

Abstract: The technology disclosed corrects inter-cluster intensity profile variation for improved base calling on a cluster-by-cluster basis. The technology disclosed accesses current intensity data and historic intensity data of a target cluster, where the current intensity data is for a current sequencing cycle and the historic intensity data is for one or more preceding sequencing cycles. A first accumulated intensity correction parameter is determined by accumulating distribution intensities measured for the target cluster at the current and preceding sequencing cycles. A second accumulated intensity correction parameter is determined by accumulating intensity errors measured for the target cluster at the current and preceding sequencing cycles. Based on the first and second accumulated intensity correction parameters, next intensity data for a next sequencing cycle is corrected to generate corrected next intensity data, which is used to base call the target cluster at the next sequencing cycle.

Abstract: A method of jointly training of a transferable feature extractor network, an ordinal regressor network, and an order classifier network in an ordinal regression unsupervised domain adaption network by providing a source of labeled source images and unlabeled target images; outputting image representations from a transferable feature extractor network by performing a minimax optimization procedure on the source of labeled source images and unlabeled target images; training a domain discriminator network, using the image representations from the transferable feature extractor network, to distinguish between source images and target images; training an ordinal regressor network using a full set of source images from the transferable feature extractor network; and training an order classifier network using a full set of source images from said transferable feature extractor network.

Abstract: According to some embodiments, a method performed by a classification scanner comprises receiving an electronic message and determining a classification that applies to the electronic message. The classification is determined based on an express indication from a user. The method further comprises providing a machine learning trainer with the electronic message and an identification of the classification that applies to the electronic message. The machine learning trainer is adapted to determine a machine learning policy that associates attributes of the electronic message with the classification.

Abstract: A model may be trained on a training dataset, e.g., for medical image processing or medical signal processing tasks. Systems and computer-implemented methods are provided for associating a population descriptor with the trained model and using the population descriptor to determine whether records to which the model is to be applied, conform to the population descriptor. The population descriptor characterizes a distribution of the one or more characteristic features over the training dataset, with the characteristic features characterizing the training record and/or a model output provided when the trained model is applied to the training record. For instance, the model may be applied only to records conforming to population descriptor, or model outputs of applying the model to non-conforming records may be flagged as possibly untrustworthy.

Abstract: Disclosed is a few-shot gesture recognition method. The method comprises the following steps: customizing, by a user, gesture categories, and acquiring few samples for each gesture category; inputting the acquired samples into a trained few-shot learning model, extracting a feature vector corresponding to each sample, and synthesizing feature vectors belonging to the same gesture to obtain an average feature vector corresponding to each gesture as a prototype vector; acquiring a corresponding sample for a target gesture implemented by the user, and inputting the sample into the few-shot learning model to obtain a feature vector of the target gesture as a query vector; and calculating similarities between the query vector and prototype vectors of different gestures, and selecting a gesture category corresponding to the prototype vector with the highest similarity as a prediction category of the target gesture.

Abstract: Object detection architectures for detecting and classifying objects in an image are modified to incorporate an extending Rapid Class Augmentation (XRCA) progressive learning algorithm with its defining aspect of memory built into its optimizer which allows joint optimization over both the old and the classes using just the new class data and eliminates the issues associated with catastrophic forgetting.