Abstract: A micro-engagement platform system and method for creating and launching micro-engagements is disclosed. The platform enables creation of a set of one or more micro-engagements, which are short snippets of information exchanged as part of an overall goal of customer engagement; couples micro-engagements to intended customers on their mobile device via messaging; collects responses that a customer sends immediately or at a later point in time; initiates the next micro-engagement in the sequence after getting a response from the customer to the previous micro-engagement as part of the overall goal. The platform chooses from one of a plurality of modes of delivery for that snippet of information; modes can be SMS, mobile browser or ore or more mobile apps.

Abstract: A data processing system and related method. The system comprises an input interface for receiving data to be data-mined for a target. A categorizer configured to compute one or more category operators suitable to categorize said data. The computing by the categorizer of the category operators includes computing a metric. The system further includes an output interface for outputting the so computed one or more category operators.

Abstract: Example implementations described herein are directed to systems and methods for feature preparation that receives patient feature data and determines similarity of pre-stored models with the patient feature data. In an example implementation, a database of the pre-stored models is analyzed to assess similarity indicating that feature preparation of the pre-stored models is compatible with the patient feature data. For similarity indicative of feature preparation to be utilized, the feature preparation is conducted for the patient feature data based on the pre-stored model determined to be similar. The feature preparation retrieves reusable features associate with the similar pre-stored model, where the reusable features comprise pre-calculated features of the model. A machine learning model is generated using results of the feature preparation and patient feature data; and a prediction is provided using the machine learning model.

Abstract: A method of processing raw data as it is received from a data provider via an input channel is disclosed. Values are derived from the raw data as it is received from the data provider via the input channel. The derived values correspond to custom fields of a summary table. The summary table is configured to store a summary of the raw data. The custom fields correspond to data capable of improving an analysis of an entity by an analysis tool. The derived values are inserted into the custom fields of the summary table. Access to the summary table is provided as enriched data for use by the analysis tool to improve the analysis of the entity.

Abstract: Predicting business outcomes for a target user includes generation of predictor models based on test data of tests users. The test data includes historical data of the test users, resumes of the test users, and answers provided by the test users to psychometric questions. The predictor models are then used to predict psychometric features and business outcomes based on target data of the target user. The target data includes a resume of the target user, historical data of the target user, and answers provided by the target user to the psychometric questions.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for generating output sequences from input sequences. One of the methods includes obtaining an input sequence having a first number of inputs arranged according to an input order; processing each input in the input sequence using an encoder recurrent neural network to generate a respective encoder hidden state for each input in the input sequence; and generating an output sequence having a second number of outputs arranged according to an output order, each output in the output sequence being selected from the inputs in the input sequence, comprising, for each position in the output order: generating a softmax output for the position using the encoder hidden states that is a pointer into the input sequence; and selecting an input from the input sequence as the output at the position using the softmax output.

Abstract: Method, electronic device, and computer readable medium embodiments are disclosed. In one embodiment, a method includes training a neural network using a first image dataset and a first truth dataset, then using the trained neural network to analyze a second image dataset. The training includes modifying a loss function of the neural network to forego penalizing the neural network when a feature is predicted with higher than a first confidence level by the neural network, and the first truth dataset has no feature corresponding to the predicted feature.

Abstract: A system and method for predicting a sequence of actions employ a Gated End-to-End Memory Policy Network (GMemN2NP), which includes a sequence of hop(s). Supporting memories of the hops include memory cells generated from observations made at different times. A sequence of actions is predicted, based on input agent-specific variables. For each action, the model, at each hop, outputs an updated controller state which is used as input to the next hop or, for the terminal hop, for computing the respective action. Each hop includes a transform gate mechanism which is used to control the influence of output of the supporting memories on the updated controller state. For the second and subsequent hops, respective actions are predicted, after using any intervening observations to update the supporting memories. The model is learned, on a training set of observations, to optimize the cumulative reward of a sequence of two or more actions.

Abstract: Techniques of forecasting web metrics involve generating, prior to the end of a period of time, a probability of a metric taking on an anomalous value, e.g., a value indicative of an anomaly with respect to web traffic, at the end of the period based on previous values of the metric. Such a probability is based on a distribution of predicted values of the metric at some previous period of time. For example, a web server may use actual values of the number of bounces collected at hourly intervals in the middle of a day to predict a number of bounces at the end of the current day. Further, the web server may also compute a confidence interval to determine whether a predicted end-of-day number of bounces may be considered anomalous. The width of the confidence interval indicates the probability that a predicted end-of-day number of bounces has an anomalous value.

Abstract: A heating, ventilation, and air conditioning (HVAC) control device is configured to record a plurality of actual occupancy statuses, to determine a plurality of corresponding predicted occupancy statuses, and to compare the plurality of predicted occupancy statuses to the plurality of actual occupancy statuses. The device is further configured to identify conflicting occupancy statuses based on the comparison. A conflicting occupancy status indicates a difference between an actual occupancy status and a corresponding predicted occupancy status. The device is further configured to identify timestamps corresponding with the conflicting occupancy statuses, to identify historical occupancy statuses corresponding with the identified timestamps, and to update the conflicting occupancy statuses in the predicted occupancy schedule with the historical occupancy statuses.

Abstract: A power generation prediction system using a first and second neural networks is provided, and the first neural network is connected to the second neural network. The first neural network receives first input data, and generates the amount prediction data according to the first input data. The first input data is used to determine amount prediction data, and the amount prediction data is used to determine power generation prediction data. The second neural network receives the amount prediction data, and calculates the power generation prediction data according to the amount prediction data. When a device in a selected area is deteriorated or reinstalled, the second neural network is fine-tuned and trained again. The power generation prediction data is a power generation prediction bound having a maximum and minimum power generation prediction values, and thus the power deployment terminal in a power grid can deploy power more precisely.

Abstract: A mechanism is provided in a data processing system comprising at least one processor and at least one memory, the at least one memory comprising instructions executed by the at least one processor to cause the at least one processor to implement a drug response estimation engine. The drug response estimation engine receives real-world evidence for a plurality of patients. A patient similarity network builder component executing within the drug response estimation engine builds a patient similarity network. A regression analysis component executing within the drug response estimation engine builds a network localized regression analysis approach. A patient clustering component executing within the drug response estimation engine groups patients based on demographics and comorbidities to form a plurality of patient clusters.

Abstract: An embodiment of the present disclosure provides an arc detection method, in which an apparatus detects arcs, comprising the steps of: obtaining time series data for measured values of an electric current flowing in a wire; calculating first statistical values indicating dispersion degrees with time of the measured values or dispersion degrees with time of variances of the measured values from the time series data; and determining that an arc occurs in the wire or that the possibility of arc occurrence in the wire is high in a case when at least one of the first statistical values is out of a predefined range.

Abstract: Some embodiments provide a method for training a machine-trained (MT) network that processes inputs using network parameters. The method propagates a set of input training items through the MT network to generate a set of output values. The set of input training items comprises multiple training items for each of multiple categories. The method identifies multiple training item groupings in the set of input training items. Each grouping includes at least two training items in a first category and at least one training item in a second category. The method calculates a value of a loss function as a summation of individual loss functions for each of the identified training item groupings. The individual loss function for each particular training item grouping is based on the output values for the training items of the grouping. The method trains the network parameters using the calculated loss function value.

Abstract: In one embodiment, a supervisory service for a software-defined wide area network (SD-WAN) obtains telemetry data from one or more edge devices in the SD-WAN. The service trains, using the telemetry data as training data, a machine learning-based model to predict tunnel failures in the SD-WAN. The service receives feedback from the one or more edge devices regarding failure predictions made by the trained machine learning-based model. The service retrains the machine learning-based model, based on the received feedback.

Abstract: The embodiments recite systems and methods that improve the traditional underwriting process within a financial institution. These embodiments produce an underwriting model that emulates the resolution patterns of top performing underwriters. The underwriting model once is built and tested is incorporated into decision tools that provide underwriters with insightful advices when underwriting a client. The embodiments use statistical learning techniques such as support vector machine and logistic regression. These techniques can assume a linear or nonlinear relationship between factors and risk classes. Furthermore, the underwriting model also uses artificial intelligence tools such as expert systems and fuzzy logic. A company's underwriting standards and best underwriting practices may be updated periodically so that underwriting model based on decision heuristic keep improving the quality of its output over time.

Abstract: A method, computer program product, and apparatus for adapting a trained neural network having one or more batch normalization layers are provided. The method includes adapting only the one or more batch normalization layers using adaptation data. The method also includes adapting the whole of the neural network having the one or more adapted batch normalization layers, using the adaptation data.

Abstract: Methods and systems described in this disclosure receive, from a device, information associated with an activity for input into an enterprise integrated calendar, including a start date of the activity, a completion date of the activity, and a resource expected to be impacted by the activity from the start date to the completion date. Based on the start and completions dates and the resource impacted, a change impact score for the activity can be determined. A saturation score (or a series of saturation scores) for the resource can be generated from the start date to the completion date based on the change impact score for the activity and change impact scores of other activities. Such scores can provide information about whether the activity should remain on the enterprise calendar or be moved to a different date where the saturation scores do not exceed certain saturation levels.

Abstract: A method for generating predicted survey participation data at a current user device. A server collects training behavioral data related to a website from a plurality of user devices. The server also collects training survey participation data related to the website from at least some of the plurality of user devices. The server analyzes the training survey participation data and the corresponding training behavioral data to infer correlations between the training survey participation data and the corresponding training behavioral data. The server further generates predictive survey participation patterns based on the inferred correlations. The server transmits the predictive survey participation patterns to the current user device. The current device collects current behavioral data related to the website.

Abstract: According to some embodiments, a hail history storage device may store information periodically received from a remote third-party weather reporting service. A hail history score request associated with a geographic location and a date range may be received from a remote requester device. Responsive to the received hail history score request, a computer processor of a hail history server may automatically access information in the hail history storage device based on the geographic location and date range. Moreover, the computer processor may automatically evaluate accessed information to calculate a hail history score value. The computer processor may then transmit, to the remote requester device, historical hail evaluation data including the calculated hail history score value.

Abstract: A method, apparatus and product for testing a system under development. The method comprises obtaining information including a current version of a code, and predicting that a subset of code elements from the current version of the code are bug-prone elements. The method comprises determining one or more coverage events corresponding to the bug-prone elements. The method further comprises determining a testing policy based on the one or more coverage events that correspond to the bug-prone elements, wherein the testing policy is determined based on a statistical analysis of coverage likelihood of tests generated based on a test template for each coverage event of the one or more coverage events. The method further comprises implementing the testing policy.

Abstract: Training and/or utilizing an interaction prediction model to generate a predicted interaction value that indicates a likelihood of interaction with a corresponding application on the basis of an electronic communication. The application can be in addition to any electronic communication application that is utilized in formulating the electronic communication and/or that is utilized in rendering the electronic communication. The predicted interaction value can be generated based on processing, utilizing the interaction prediction model, of features of the electronic communication and/or of other features. The predicted interaction value can be utilized to determine whether to perform further action(s) that interact with, and/or enable efficient interaction with, the application on the basis of the electronic communication.

Abstract: A mechanism is provided in a data processing system comprising at least one processor and at least one memory, the at least one memory comprising instructions executed by the at least one processor to cause the at least one processor to implement a drug response estimation engine. The drug response estimation engine receives real-world evidence for a plurality of patients. A patient similarity network builder component executing within the drug response estimation engine builds a patient similarity network. A regression analysis component executing within the drug response estimation engine builds a network localized regression analysis approach. A patient clustering component executing within the drug response estimation engine groups patients based on demographics and comorbidities to form a plurality of patient clusters.

Abstract: A graphical structure model trained with labeled samples is obtained. The graphical structure model is defined based on an account relationship network that comprises a plurality of nodes and edges. The edges correspond to relationships between adjacent nodes. Each labeled sample comprises a label indicating whether a corresponding node is an abnormal node. The graphical structure model is configured to iteratively calculate, for at least one node of the plurality of nodes, an embedding vector in a hidden feature space based on an original feature of the least one node and/or a feature of an edge associated with the at least one node. A first embedding vector that corresponds to a to-be-tested sample is calculated using the graphical structure model. Abnormal account prevention and control is performed on the to-be-tested sample based on the first embedding vector.

Abstract: As part of neural network sensitivity analyses, base outputs of hidden layer nodes of a neural network model for non-perturbed variables can be reused when perturbing the variables. Such an arrangement greatly reduces complexity of the calculations required to generate outputs of the model. Related apparatus, systems, techniques and articles are also described.

Abstract: In exemplary aspects of optimizing data centers, historical data corresponding to a data center is collected. The data center includes a plurality of systems. A data center representation is generated. The data center representation can be one or more of a schematic and a collection of data from among the historical data. The data center representation is encoded into a neural network model. The neural network model is trained using at least a portion of the historical data. The trained model is deployed using a first set of inputs, causing the model to generate one or more output values for managing or optimizing the data center using supplemental indicators.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for obtaining a plurality of model representations of predictive models, each model representation associated with a respective user and expresses a respective predictive model, and selecting a model implementation for each of the model representations based on one or more system usage properties associated with the user associated with the corresponding model representation.

Abstract: Implementations are directed to receiving current data, processing the current data using a predictive model to provide a result, the result corresponding to a sub-model of the predictive model, determining a set of syntactically similar sub-models based on other data, providing at least one semantic model based on the sub-model of the predictive model, one or more syntactically similar sub-models of the set of syntactically similar sub-models, a domain ontology (knowledge graph), and constraints, the at least one semantic model being provided by merging nodes of the sub-model of the predictive model, and a previously determined sub-model of the predictive model using the domain ontology, a label of the domain ontology being used to label a merged node, determining an interpretation based on the at least one semantic model, the interpretation providing at least one reason for the result, and providing the interpretation.

Abstract: Disclosed are an encoding device and a decoding device, which include a CNN-based in-loop filter. The encoding device according to one embodiment comprises: a filtering unit for generating filtering information by filtering a residual image corresponding to a difference between an original image and a prediction image; an inverse filtering unit for generating inverse filtering information by inversely filtering the filtering information; a prediction unit for generating the prediction image on the basis of the original image and reconstruction information; a CNN-based in-loop filter for receiving the inverse filtering information and the prediction image so as to output the reconstruction information; and an encoding unit for performing encoding on the basis of the filtering information and information of the prediction image.

Abstract: Methods and apparatus are provided for process monitoring based on large-scale combinations of time series data. An exemplary method comprises generating a model from time series data for a given target time series; determining whether a first difference between measured values and predicted values based on the model exceeds a predefined threshold indicating a target prediction error; in response to a detected target prediction error, performing evaluations of (i) a neighborhood coherence comprising an average of variables of the model weighted by corresponding coefficients on a predefined neighborhood time window, and/or (ii) a second difference between a given value of at least one variable in the model and an average value of the at least one variable based on a training dataset; providing notifications when first predefined criteria based on the evaluations are satisfied; and updating the model when second predefined criteria based on the evaluations are satisfied.

Abstract: Particular embodiments may receive, by one or more processors of an information processing system, results of a benchmark task performed by a set of one or more workers. Performance of each of the workers in the set on the benchmark task may be determined. One or more best workers of the workers in the set may be selected. Particular embodiments may then determine whether at least one of the best workers is available for a new task. If one of the best workers is available, the new task may be assigned to the available best worker; else the new task may be assigned to a random worker.

Abstract: A system and method for crowd-sourced refinement of natural phenomenon for risk management and contract validation, comprising at least a heterogeneous mixture of sensors and data-gathering techniques, a sensor fusion suite, and a business operating system, which ingests, transforms if necessary, and analyzes received data and develops and applies models of prediction of consequences of the sensor data and future events based on such data for purposes such as insurance liability and risk assessment, emergency services planning, and financial market predictions, and comparing historical models and data with current data and models to attempt to refine and utilize a more precise predictive model for these purposes.

Abstract: Systems and methods are provided for identifying a safe window for creating a firmware update package. In some embodiments, an Information Handling System (IHS), may include a processor and a memory coupled to the processor, the memory having program instructions stored thereon that, upon execution, cause the IHS to: generate tokens based upon a log of historical firmware update package creation events; identify a plurality of workflows based upon the tokens; predict, based upon the plurality of workflows, a future workload of a firmware update package creation service; and indicate, based upon the future workload, a time window for creating a firmware update package using the firmware update package creation service that increases a probability of successful creation.

Abstract: Computing apparatus includes a central processing unit (CPU), including at least one core and a cache in physical proximity to the at least one core, with a system memory and a bus connecting the CPU to the memory. A peripheral device is connected to the bus and is configured to write data items via the bus to a buffer in the system memory and to write respective completion reports to the system memory upon writing the data items to the buffer. The peripheral device is configured to detect that the CPU has read a first completion report from the system memory and then read context metadata associated with the first completion report from a given address in the system memory, and is further configured, upon writing a second completion report subsequent to the first completion report and associated with the same context metadata, to stash the second completion report and the context metadata from the given address to the cache.

Abstract: A method in one embodiment comprises defining a plurality of fields in a plurality of electronic documents, wherein the plurality of fields respectively correspond to governance, risk and compliance system data structures, identifying a plurality of relationships between the electronic documents based on one or more cross-references between fields of two or more different electronic documents of the plurality of electronic documents, and assigning respective ranks to the plurality of electronic documents based on the relationships. In the method, a query is received from a user device, and a listing of candidate documents of the plurality of electronic documents is retrieved in response to the query. Scores for respective ones of the candidate documents are computed based on at least the assigned ranks, and a response to the query is transmitted to the user device, wherein the response comprises the listing of candidate documents sorted according to the computed scores.

Abstract: At an electronic device with a display screen, display a message transcript, where the message transcript includes an incoming message. Determine, based at least in-part on the message, a plurality of suggested one or more characters. Determine if the user of the device has, in the past, frequently inputted a response different from the plurality suggested one or more characters when the plurality were presented for user selection. Optionally, determine if the frequently inputted response is synonymous with a suggested one or more characters. Display the frequently-inputted response, in place of at least one of the plurality of suggested one or more characters, under some circumstances based on these determination(s).

Abstract: First FEA mesh model representing 3-D geometry of a carbon fiber reinforced composite (CFRC) product/part, pre-forming fiber orientation and desired reference fiber direction at a particular location on the product/part are received. First FEA mesh model contains finite elements associated with respective material properties for carbon fibers and binding matrix. Pre-forming fiber orientation includes number of fibers and relative angles amongst the fibers. Pre-forming 2-D shape of a workpiece used for manufacturing the product/part is obtained by conducting a one-step inverse numerical simulation that numerically expands the first to a second FEA mesh model based on numerically-calculated structural behaviors according to respective material properties. Pre-forming fiber orientation is superimposed on the second FEA mesh model with the desired reference fiber direction being preserved.

Abstract: The present invention provides a method for monitoring effectiveness of a three-way catalytic converter of a gasoline engine. The method includes detecting emissions of the gasoline engine by an NOX sensor arranged within the three-way catalytic converter or downstream of the three-way catalytic converter in an exhaust-gas flow direction and transmitting values measured by the NOX sensor together with operating variables of the gasoline engine and/or of the three-way catalytic converter to a processor. The processor is configured to execute a simulation model and thereby calculate emissions values of a reference catalytic converter on the basis of the operating variables. The method further includes comparing the calculated emissions values with the values measured by the NOX sensor.

Abstract: Entity resolution (i.e., record linkage) involves the analysis/discovering of datasets that refer to the same real world entity. Analysis typically involves transformation and comparison of different fields of the dataset followed by the application of often domain/data specific logic for determining datasets that refer to the same real world entity (e.g., person). Consider, a bulk mailing of product catalogs to potential customers. Some individuals may have numerous public records that identify the individual differently. Illustratively, several records associated with Jane Doe at her current home address may exist: one record with her name listed as J. Doe, a second record as Jane H. Doe, a third record as Doe, Jane, and a fourth record as Jan Doe (a misspelling). Conceivably, the bulk mailing could unwittingly send multiple catalogs to Jane Doe at her current address, one for each name variation. The entity resolution process described herein can overcome such problems.

Abstract: Systems and methods for training models to improve fairness.

Abstract: Tiered advertisement bidding is disclosed. One or more quality metrics associated with a user profile are determined. An advertisement bid is selected from a plurality of tiered bids based at least in part on the determined quality metrics. Determining the quality metrics can include determining a conversion assessment. Determining the quality metric can also include determining whether a need associated with a user profile has been met for a category. In some cases, persona detection is performed with respect to the user profile.

Abstract: A method for searching a point of interest (POI) around a location is provided. The method includes receiving information relating to a target location associated with a user from a terminal and determining a target geocode value relating to the target location. The method also includes obtaining a geocode structure for determining POIs. The method further includes determining at least one POI relating to the target location based on the geocode structure and the target geocode value.

Abstract: Methods and systems described herein are directed towards using artificial intelligence to recommend network features to add to networks. A network configuration system may determine the network components and usage characteristics for a network within a plurality of networks. The network configuration system may use an ensemble to generate recommendations based on the network components and usage characteristics of the plurality of networks. The ensemble may include one or more machine learning based recommenders. The ensemble may also include a non-personalized recommender that generates recommendations based on known compatibility between network features. The network configuration system may weight recommendations from each recommender to create a ranked ordering of the recommendations. The network configuration system may modify (e.g., add or remove network features) one or more networks based on the recommendations and/or user input.

Abstract: A method can include receiving sensor data from at least three different types of sensor situated in the geographic area, the types of sensors including an air temperature sensor, relative humidity sensor, dewpoint sensor, soil moisture sensor, soil temperature sensor, average wind speed sensor, maximum wind speed sensor, and a rainfall sensor, producing a feature vector including a time series of values corresponding to the received sensor data, and using a neural network, estimating the physical characteristics, the physical characteristics including at least one of (a) a leaf wetness, (b) a solar radiation, (c) an evapotranspiration, (d) a future soil moisture, and (e) a future soil temperature.

Abstract: We generally describe a system (100) for antenna beam selection in a radio access network. The system (100) comprises a measurement module (102) which is configured to obtain channel quality measurements of a plurality of beams usable to serve a user equipment (112) in the 5G radio access network. The system (100) further comprises a training module (104) coupled to the measurement module (102), wherein the training module (104) is configured to generate a machine learning model based on the channel quality measurements. The system (100) further comprises a prediction module (106) coupled to the training module (104), wherein the prediction module (106) is configured to receive the machine learning model from the training module (104), and select, based on the machine learning model, one of the plurality of beams used to serve the user equipment (112).

Abstract: Disclosed herein is a system and method that can be used with any underlying classification technique. The method receives a test dataset and determines the features in that test dataset that are present. From these features the training dataset is modified to only have those features that are present in the test dataset. This modified test dataset is then used to calibrate the classifier for the particular incoming data set. The process repeats itself for each different incoming dataset providing a just in time calibration of the classifier.

Abstract: Systems, methods, and devices for evaluation of an earth formation intersected by a borehole using a logging tool. Methods include making EM measurements on a single logging run while drilling using an EM tool on a tool string in a substantially horizontally aligned section of the borehole at a plurality of borehole depths; estimating, at each of the plurality of borehole depths, a true resistivity of a volume of interest of the formation in which the tool sits in substantially real time while on the single logging run while drilling.

Abstract: Techniques are described for generating predictive alerts. In one or more embodiments, a seasonal model is generated, the seasonal model representing one or more seasonal patterns within a first set of time-series data, the first set of time-series data comprising data points from a first range of time. A trend-based model is also generated to represent trending patterns within a second set of time-series data comprising data points from a second range of time that is different than the first range of time. A set of forecasted values is generated based on the seasonal model and the trend-based model. Responsive to determining that a set of alerting thresholds has been satisfied based on the set of forecasted values, an alert is generated.

Abstract: Disclosed are a method, computer program and a metrology apparatus for measuring a process effect parameter relating to a manufacturing process for manufacturing integrated circuits on a substrate. The method comprises determining for a structure, a first quality metric value for a quality metric from a plurality of measurement values each relating to a different measurement condition while cancelling or mitigating for the effect of the process effect parameter on the plurality of measurement values and a second quality metric value for the quality metric from at least one measurement value relating to at least one measurement condition without cancelling or mitigating for the effect of the process effect parameter on the at least one measurement value. The process effect parameter value for the process effect parameter can then be calculated from the first quality metric value and the second quality metric value, for example by calculating their difference.

Abstract: System and method embodiments are provided for adaptive anomaly detection based predictor for network data. In an embodiment, a computer-implemented method in a network component for predicting values of future network time series data includes receiving, with one or more receivers, network time series data; determining, with one or more processors, whether an anomaly is detected in the network time series data; generating, with the one or more processors, a prediction associated with the network data according to a primary predictor when no anomaly is detected in the network time series data; generating, with the one or more processors, the prediction associated with the network data according to an alternative predictor when an anomaly in the network time series data is detected; and sending, with one or more transmitters, the prediction to a network controller, wherein the network controller uses the prediction to adjust network parameters.