Abstract: Provided is a command inspection method, including: inspecting a plurality of to-be-inspected commands based on a plurality of first features of known obfuscated commands and a plurality of second features of normal service commands, to determine a plurality of target commands, each of the target commands being a command having one or more of the first features but none of the second features; obtaining running state text of the plurality of target commands; obtaining suspicious degree information of the plurality of target commands based on the plurality of target commands and the corresponding running state text, the suspicious degree information being used for indicating a possibility that a command is an obfuscated command; and determining, when suspicious degree information of any target command satisfies a target condition, the target command as an obfuscated command.

Abstract: A calculation device, a calculation method, and a storage medium are provided. The present invention is provided with: a first calculation unit which, with respect to input data, performs calculations relating to predetermined processing by using a first model in which a corresponding relationship between input data and output data changes by performing machine learning that uses learning data, and outputs a first output; a second calculation unit which, with respect to input data, performs calculations relating to predetermined processing using a second model for which the correspondence relationship between input data and output data is fixed, and outputs a second output; and a comparison unit which, on the basis of a comparison result obtained by comparing the first output and the second output with a prescribed determination standard, outputs the first output, the second output, or a third output that is a combination of the first and second outputs.

Abstract: The apparatus employs adaptive machine learning for model selection based on data complexity and user goals. It consists of a processor and memory. Initially, it creates a first model from a dataset and analytic goals. Then, it determines a complexity metric for another dataset, compares this metric to a set threshold, and identifies a complexity gap. Using a feature learning algorithm, it extracts candidate features from the second dataset. From these features, it generates a second model. The device assesses this model's performance using a third dataset and selects it based on its relation to the complexity gap.

Abstract: Embodiments are disclosed for a quality control method. The method includes verifying one or more sensors as part of a verification process and assessing a quality of an item or a technical process as part of a quality control process. Two or more measurements values characterizing the item or the technical process from the one or more sensors are obtained. One or more output values based on a computerized process taking the two or more measurements values as inputs are obtained. At least one of the output values obtained is compared to one or more corresponding reference values, to obtain a quality assessment of the item or the technical process.

Abstract: Systems and methods enable automated and scalable obfuscation detection in programming scripts, including processing devices that receive software programming scripts and a symbol set. The processing devices determine a frequency of each symbol and an average frequency of the symbols in the script text. The processing devices determine a normal score of each symbol based on the frequency of each symbol and the average frequency to create a symbol feature for each symbol including the normal score. The processing devices utilize an obfuscation machine learning model including a classifier for binary obfuscation classification to detect obfuscation in the script based on the symbol features. The processing devices cause to display an alert indicting an obfuscated software programming script on a screen of a computing device associated with an administrative user to recommend security analysis of the software programming script based on the binary obfuscation classification.

Abstract: A method is used to design nuclear reactors using design variables and metric variables. A user specifies ranges for the design variables and target values for the metric variables. A set of design parameter samples are selected. For each sample, the method runs three processes, which compute metric variables to thermal-hydraulics, neutronics, and stress. The method applies a cost function to each sample to compute an aggregate residual of the metric variables compared to the target values. The method trains a machine learning model using the samples and the computed aggregate residuals. The method shrinks the range for each design variable according to correlation between the respective design variable and estimated residuals using the machine learning model. These steps are repeated until a sample having a smallest residual is unchanged for multiple iterations. The method then uses the final machine learning model to assess relative importance of each design variable.

Abstract: A system and method for configuring an RF network based on machine learning. In some embodiments, the method includes: receiving, by a first neural network, a first state and a first state transition, the first state including: one or more identifiers for available active ports, and a set of available connections between two or more circuit elements, each of the circuit elements being one of: (1) a first circuit type, (2) a second circuit type that operatively connects a circuit element of the first circuit type to one of the available active ports, and (3) the available active ports; and generating, by the first neural network, a first estimated quality value, for the first state transition.

Abstract: In some implementations, a system may receive a shell script associated with a computing device. The system may generate a character frequency feature vector based on the shell script. The system may input text of the shell script to a convolutional neural network (CNN) branch of a trained deep learning model. The system may input the character frequency feature vector to a feedforward neural network (FNN) branch of the trained deep learning model. The system may determine using the trained deep learning model, a respective probability score for each of a plurality of obfuscation types for the shell script based on a combined output of the CNN branch and the FNN branch. The system may detect whether the shell script is obfuscated based on the respective probability score for each of the plurality of obfuscation types determined for the shell script.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training an action selection neural network that is used to select actions to be performed by a reinforcement learning agent interacting with an environment. In particular, the actions are selected from a continuous action space and the system trains the action selection neural network jointly with a distribution Q network that is used to update the parameters of the action selection neural network.

Abstract: Systems and methods are described for facilitating operation of a plurality of computing devices. Data indicative of enumerated resources of a computing device is collected. The data is collected without dependency on write permissions to a file system of the one computing device. A condition of the computing device is determined based on historical data associated with enumerated resources of other computing devices. The identified condition can be updated as updated historical data becomes available. A communication to the computing device may be sent based on the identified condition.

Abstract: This disclosure relates generally to system and method for forecasting location of target in monocular first person view. Conventional systems for location forecasting utilizes complex neural networks and hence are computationally intensive and requires high compute power. The disclosed system includes an efficient and light-weight RNN based network model for predicting motion of targets in first person monocular videos. The network model includes an auto-encoder in the encoding phase and a regularizing layer in the end helps us get better accuracy. The disclosed method relies entirely just on detection bounding boxes for prediction as well as training of the network model and is still capable of transferring zero-shot on a different dataset.

Abstract: A computer-implemented method is provided for determining an action with respect to a given portfolio of items for supply chain management. The method includes acquiring, by a hardware processor, a feature vector for supply chain delivery trends, the given portfolio, and a current investment amount. The method further includes determining, by the hardware processor, whether a current supply chain delivery situation is normal or abnormal based on the feature vector. The method also includes performing a risk-avoidance action to reduce the current investment amount and avoid potential supply chain delivery losses, responsive to a determination that the current supply chain delivery situation is abnormal. The method additionally includes performing a risk adaptive action to increase the current investment amount and incur potential supply chain delivery gains by using a distributional reinforcement learning process, responsive to a determination that the current supply chain delivery situation is normal.

Abstract: According to one or more embodiments of the present invention, a computer-implemented method for adjusting a process variable using a closed loop system includes initializing a radial basis function neural network (RBF network) using a maximum error (emax), a maximum first order change in error (?emax), a maximum second order change in error (?2emax), and a maximum output increment (?omax), associated with the closed loop system being controlled. The method further includes inputting, to the RBF network, input values including an error, a first order change in error, and a second order change in error. The method further includes computing, by the RBF network, control parameters based on the input values, and computing, by the RBF network, an incremental change in the process variable based on the control parameters. The method further includes adjusting, by a controller, an output device to change the process variable by the incremental change.

Abstract: Described herein is a method and system for controlling an interior-mounted permanent magnet (IPM) alternating-current (AC) electrical machine utilizing a space vector pulse-width modulated (SVPWM) converter operably connected between an electrical power source and the IPM AC electrical machine comprising three neural networks (NNs), including a controller NN operably connected to the SVPWM converter, a parameter estimator NN, and a flux-weakening and MTPA NN.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training an action selection neural network used to select actions to be performed by an agent interacting with an environment. In one aspect, a system comprises a plurality of actor computing units and a plurality of learner computing units. The actor computing units generate experience tuple trajectories that are used by the learner computing units to update learner action selection neural network parameters using a reinforcement learning technique. The reinforcement learning technique may be an off-policy actor critic reinforcement learning technique.

Abstract: A system provides feedback driven end-to-end state control of a data model. A data model may be used to model the behavior of a petrochemical refinery to predict future events. The system may be used to ensure proper operation of the data model. Contingency data models may be executed when a failure is detected. Further, when the system detects accuracy that is out of tolerance, the system may initiate retraining of the data model being currently used.

Abstract: Described is a solution for an unlabeled target domain dataset challenge using a domain adaptation technique to train a neural network using an iterative 3D model fitting algorithm to generate refined target domain labels. The neural network supports the convergence of the 3D model fitting algorithm and the 3D model fitting algorithm provides refined labels that are used for training of the neural network. During real-time inference, only the trained neural network is required. A convolutional neural network (CNN) is trained using labeled synthetic frames (source domain) with unlabeled real depth frames (target domain). The CNN initializes an offline iterative 3D model fitting algorithm capable of accurately labeling the hand pose in real depth frames. The labeled real depth frames are used to continue training the CNN thereby improving accuracy beyond that achievable by using only unlabeled real depth frames for domain adaptation.

Abstract: Methods, devices, systems, and instructions for adaptive quantization in an artificial neural network (ANN) calculate a distribution of ANN information; select a quantization function from a set of quantization functions based on the distribution; apply the quantization function to the ANN information to generate quantized ANN information; load the quantized ANN information into the ANN; and generate an output based on the quantized ANN information. Some examples recalculate the distribution of ANN information and reselect the quantization function from the set of quantization functions based on the resampled distribution if the output does not sufficiently correlate with a known correct output. In some examples, the ANN information includes a set of training data. In some examples, the ANN information includes a plurality of link weights.

Abstract: The present disclosure provides techniques for graphics translation. A plurality of natural language image descriptions is collected for an image of a product. An overall description for the image is generated using one or more models, based on the plurality of natural language image descriptions, by: identifying a set of shared descriptors used in at least a subset of the plurality of natural language image descriptions, and aggregating the set of shared descriptors to form the overall description. A first request to provide a description of the first image is received, and the overall description is returned in response to the first request, where the overall description is output using one or more text-to-speech techniques.

Abstract: A substrate treating apparatus and a substrate treating method are disclosed. The substrate treating apparatus includes a process module configured to perform processing on a substrate, an index module configured to insert the substrate into the process module and withdraw the substrate, of which the processing is completed, from the process module, a loading module configured to relay the substrate between the process module in a vacuum atmosphere and the index module in an atmospheric pressure atmosphere by switching an atmosphere thereof to the vacuum atmosphere or the atmospheric pressure atmosphere, and a control module configured to receive operation states from the process module, the index module, and the loading module and schedule operations of the process module, the index module, and the loading module in a direction in which the number of substrates to be processed per unit time increases with reference to the received operation states.

Abstract: Deep Learning is a candidate for advanced process control, but requires a significant amount of process data not normally available from regular plant operation data. Embodiments disclosed herein are directed to solving this issue. One example embodiment is a method for creating a Deep Learning based model predictive controller for an industrial process. The example method includes creating a linear dynamic model of the industrial process, and based on the linear dynamic model, creating a linear model predictive controller to control and perturb the industrial process. The linear model predictive controller is employed in the industrial process and data is collected during execution of the industrial process. The example method further includes training a Deep Learning model of the industrial process based on the data collected using the linear model predictive controller, and based on the Deep Learning model, creating a Deep Learning model predictive controller to control the industrial process.

Abstract: A method for active learning using sparse training data can include training a machine learning model using less than ten first training data points to generate a candidate machine learning model. The method can include performing a Monte Carlo process to sample one or more first outputs of the candidate machine learning model. The method can include testing the one or more first outputs to determine if each of the one or more first outputs satisfy a respective convergence condition. The method can include, responsive to at least one first output not satisfying the respective convergence condition, training the candidate machine learning model using at least one second training data point corresponding to the at least one first output. The method can include, responsive to the one or more first outputs each satisfying the respective convergence condition, outputting the candidate machine learning model.

Abstract: Among other things, techniques are described for steering angle calibration. An autonomous vehicle receives a steering angle measurement and a yaw rate measurement, and estimates a steering angle offset using the steering angle measurement, the yaw rate measurement, and a wheel base of the autonomous vehicle. An estimated yaw rate is determined based on a yaw rate model, the steering angle measurement and the estimated steering angle offset. The yaw rate measurement and the estimated yaw rate are compared and an action is initiated on the autonomous vehicle in response to the comparing.

Abstract: A method can include receiving sensor data from a system; encoding the sensor data to a latent space representation via a trained encoder; generating a control action using the latent space representation; and issuing an instruction that corresponds to the control action for control of the system.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training an action selection neural network that is used to select actions to be performed by a reinforcement learning agent interacting with an environment. In particular, the actions are selected from a continuous action space and the system trains the action selection neural network jointly with a distribution Q network that is used to update the parameters of the action selection neural network.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for selecting actions to be performed by an agent interacting with an environment. In one aspect, a system comprises a grid cell neural network and an action selection neural network. The grid cell network is configured to: receive an input comprising data characterizing a velocity of the agent; process the input to generate a grid cell representation; and process the grid cell representation to generate an estimate of a position of the agent in the environment; the action selection neural network is configured to: receive an input comprising a grid cell representation and an observation characterizing a state of the environment; and process the input to generate an action selection network output.

Abstract: A method for automatically generating a neural network architecture includes: loading a first computer-readable representation of a first neural network architecture including first genes representing parameters of the first neural network architecture; generating a first neural network including neurons in accordance with the first genes; deploying the first neural network in a robotic controller; training the first neural network by supplying inputs to an input processing unit connected to the first neural network and receiving outputs from an output processing unit connected to the first neural network, the training including updating synaptic weights of connections between the neurons based on responses to the inputs supplied to the input processing unit; evaluating a performance of the first neural network architecture; and generating, by the computer system, an updated computer-readable representation of an updated neural network architecture based on the evaluation of the performance the first neural n

Abstract: A style-based generative network architecture enables scale-specific control of synthesized output data, such as images. During training, the style-based generative neural network (generator neural network) includes a mapping network and a synthesis network. During prediction, the mapping network may be omitted, replicated, or evaluated several times. The synthesis network may be used to generate highly varied, high-quality output data with a wide variety of attributes. For example, when used to generate images of people's faces, the attributes that may vary are age, ethnicity, camera viewpoint, pose, face shape, eyeglasses, colors (eyes, hair, etc.), hair style, lighting, background, etc. Depending on the task, generated output data may include images, audio, video, three-dimensional (3D) objects, text, etc.

Abstract: [Problem] To prevent sight of an observation target from being lost from a visual field of a monitor. [Solution] A surgical arm system includes: an articulated arm in which a plurality of joints is rotatably connected by a plurality of links and which is capable of supporting an oblique-viewing endoscope at a tip; and a control system which controls the articulated arm to change a position and a posture of the oblique-viewing endoscope. The control system controls at least one of a rotation speed and a movement speed of the oblique-viewing endoscope in a visual field imaged through the oblique-viewing endoscope based on a position of the observation target in the visual field.

Abstract: A packaging system for packing a product includes at least one of a bagging unit, a pallet loading device, and a pallet securing device. The packaging system also includes a device to determine information regarding future events associated with the packing of the product, and to render the information available.

Abstract: Computing device and method using machine learning to optimize operations of a processing chain of a food factory. The computing device collects data representative of characteristics of a product processed by the processing chain. At least some of the collected data are received from one or more sensor monitoring operations of the processing chain. The computing device determines at least one product characteristic value based on the collected data. The computing device executes the machine learning inference engine, which uses a predictive model for inferring command(s) for controlling processing appliance(s) of the processing chain based on inputs. The inputs comprise the at least one product characteristic value. The computing device transmits the command(s) to the processing appliance(s) of the processing chain. Examples of product characteristic values comprise: a product temperature, a product humidity level, a product geometric characteristic, a product weight, and a product defect measurement.

Abstract: A system and method include receiving target image data associated with a target coating. A color model and a local color model are used to predict color differences between the target coating and a sample coating. The color model and local color model includes a feature extraction analysis process that determines image features by analyzing target pixel feature differences within the target coating. Performing an optimization routine upon the color differences for determining automotive paint components for spraying a substrate.

Abstract: Techniques for generating simulations for evaluating a performance of a controller of an autonomous vehicle are described. A computing system may evaluate the performance of the controller to navigate the simulation and respond to actions of one or more objects (e.g., other vehicles, bicyclists, pedestrians, etc.) in a simulation. Actions of the objects in the simulation may be controlled by the computing system (e.g., by an artificial intelligence) and/or one or more users inputting object controls, such as via a user interface. The computing system may calculate performance metrics associated with the actions performed by the vehicle in the simulation as directed by the autonomous controller. The computing system may utilize the performance metrics to verify parameters of the autonomous controller (e.g., validate the autonomous controller) and/or to train the autonomous controller utilizing machine learning techniques to bias toward preferred actions.

Abstract: An apparatus for driving a motor comprising a first plurality of neurons of neural network circuitry, motor circuitry, and a second plurality of neurons of the neural network circuitry. The first plurality of neurons is configured to generate a first cycle value based on a target speed. The motor circuitry is configured to control, based on the first cycle value, a set of switching elements to drive the motor. The second plurality of neurons is configured to train the second plurality of neurons to generate, based on a resulting speed value for the motor that occurs when the motor circuitry has controlled the set of switching elements to drive the motor based on the first cycle value, a second cycle value to minimize a difference between the second cycle value and the first cycle value.

Abstract: Techniques and systems for critical dimension metrology are disclosed. Critical parameters can be constrained with at least one floating parameter and one or more weight coefficients. A neural network is trained to use a model that includes a Jacobian matrix. During training, at least one of the weight coefficients is adjusted, a regression is performed on reference spectra, and a root-mean-square error between the critical parameters and the reference spectra is determined. The training may be repeated until the root-mean-square error is less than a convergence threshold.

Abstract: A model aggregation device includes a communication device able to communicate with a plurality of vehicles in which neural network models are learned, a storage device storing a part of the neural network models sent from the plurality of vehicles, and a control device. The neural network model outputs at least one output parameter from a plurality of input parameters. The control device is configured to, if receiving a new neural network model from one vehicle among the plurality of vehicles through the communication device, compare ranges of the plurality of input parameters which were used for learning the new neural network model and ranges of the plurality of input parameters which were used for learning a current neural network model stored in the storage device to thereby determine whether to replace the current neural network model with the new neural network model.

Abstract: An artificial intelligence system and method for state-based learning using one or more adaptive response states of the artificial intelligence system are provided. A controller for modifying a neural network engine is configured to monitor a data stream having a data pattern by comparing the data pattern to a trained data pattern; identify a change in the data pattern of the data stream; determine a response state of the neural network learning engine, the state defining one or more neural network parameters for monitoring the data stream with the neural network learning engine; identify a predetermined policy for reconfiguring the neural network learning engine based on the data pattern and the response state; and in response to identifying the change in the data pattern and determining the response state, reconfigure the one or more neural network parameters according to the predetermined policy.

Abstract: An automated robot design pipeline facilitates the overall process of designing robots that perform various desired behaviors. The disclosed pipeline includes four stages. In the first stage, a generative engine samples a design space to generate a large number of robot designs. In the second stage, a metric engine generates behavioral metrics indicating a degree to which each robot design performs the desired behaviors. In the third stage, a mapping engine generates a behavior predictor that can predict the behavioral metrics for any given robot design. In the fourth stage, a design engine generates a graphical user interface (GUI) that guides the user in performing behavior-driven design of a robot. One advantage of the disclosed approach is that the user need not have specialized skills in either graphic design or programming to generate designs for robots that perform specific behaviors or express various emotions.

Abstract: The present disclosure provides an output method for multiple neural networks. The method includes dividing an operator operation process for each of the neural networks or operator operation processes for part of the neural networks into multiple times of executions according to a preset ratio of output frame rates among the multiple neural networks; and executing the operator operation processes for the multiple neural networks sequentially by switching among the networks, such that the multiple neural networks output uniformly and satisfy the preset ratio of output frame rates.

Abstract: A control device includes: an input information storage unit that stores track record input information that is information regarding an input signal having a track record; an artificial intelligence control unit that controls a control target using artificial intelligence based on the input signal; and an input signal evaluation unit that judges whether a value of the input signal is within a range having a track record based on the track record input information and, if the value of the input signal is within the range having the track record, permits transmission of the input signal to the artificial intelligence control unit. The safety of the control device using artificial intelligence is enhanced.

Abstract: Embodiments of the invention provide a method and system for machine failure prediction. The method comprises: identifying a plurality of basic memory depth values based on a composite sequence of machine failure history; ascertaining weight values for at least one of the identified basic memory depth values according to a pre-stored table which includes a plurality of mappings wherein each mapping relates a basic memory depth value to one set of weight values; and predicting a future failure using a Back Propagation Through Time (BPTT) trained Recurrent Neural Network (RNN) based on the ascertained weight values, wherein weight values related to a first basic memory depth value in the pre-stored table is ascertained based on a second set of weight values related to a second basic memory depth value which is less than the first basic memory depth value by a predetermined value.

Abstract: A processing system including at least one processor may obtain operational data from a radio access network (RAN), format the operational data into state information and reward information for a reinforcement learning agent (RLA), processing the state information and the reward information via the RLA, where the RLA comprises a plurality of sub-agents, each comprising a respective neural network, each of the neural networks encoding a respective policy for selecting at least one setting of at least one parameter of the RAN to increase a respective predicted reward in accordance with the state information, and where each neural network is updated in accordance with the reward information. The processing system may further determine settings for parameters of the RAN via the RLA, where the RLA determines the settings in accordance with selections for the settings via the plurality of sub-agents, and apply the plurality of settings to the RAN.

Abstract: A face anti-counterfeiting detection method includes: obtaining an image or video to be detected containing a face; extracting a feature of the image or video to be detected, and detecting whether the extracted feature contains counterfeited face clue information; and determining whether the face passes the face anti-counterfeiting detection according to a detection result.

Abstract: Methods, apparatus, systems and articles of manufacture are disclosed methods and apparatus for audio equalization based on variant selection. An example apparatus includes a processor to obtain training data, the training data including a plurality of reference audio signals each associated with a variant of music and organize the training data into a plurality of entries based on the plurality of reference audio signals, a training model executor to execute a neural network model using the training data, and a model trainer to train the neural network model by updating at least one weight corresponding to one of the entries in the training data when the neural network model does not satisfy a training threshold.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training an action selection neural network that is used to select actions to be performed by a reinforcement learning agent interacting with an environment. In particular, the actions are selected from a continuous action space and the system trains the action selection neural network jointly with a distribution Q network that is used to update the parameters of the action selection neural network.

Abstract: Provided herein is a system for occupiable space automation using neural networks that delivers scalable and more intelligent occupiable space automation that can continuously learn from user actions and experiences and adapt to specific needs of each individual occupiable space. The occupiable space automation control system is built based on brain inspired multi-layer neural network with plastic connectivity between neurons. The occupiable space automation control system is configured to (a) adaptively predict previously learned activity patterns and (b) alert about potentially harmful or undesired activity patterns of the plurality of periphery devices based on response events of the plurality of artificial neurons and coupling strengths of the plurality of synapses. The occupiable space automation control system is configured to automatically operate the at least one controller based on the predicted activity pattern and/or provide user alerts based on a detected harmful activity pattern.

Abstract: Techniques are disclosed for extracting micro-features at a pixel-level based on characteristics of one or more images. Importantly, the extraction is unsupervised, i.e., performed independent of any training data that defines particularly objects, allowing a behavior-recognition system to forgo a training phase and for object classification to proceed without being constrained by specify object definitions. A micro-feature extractor that does not require training data is adaptive and self-trains while performing the extraction. The extracted micro-features are represented as a micro-feature vector that may be input to a micro-classifier which groups object into object type clusters based on the micro-feature vectors.

Abstract: A system and method for adjusting a device personality profile. The method includes: generating a first user profile; selecting a first personality profile for a device based on matching the first user profile to a cluster of user profiles, wherein the first personality profile includes a first distribution of character traits in a multidimensional space; sampling a character trait; causing the device to execute at least a physical interaction item, based on the sampled character traits; collecting a set of real-time data from the user indicative of a reaction to the executed physical interaction item; determining a second personality profile for the device including a second distribution over a plurality of second character traits, wherein the second personality profile replaces the first personality profile; and, updating the second personality to enable execution of at least one modified physical interaction item.

Abstract: An apparatus for driving a motor includes a plurality of neurons of neural network circuitry and motor circuitry. The plurality of neurons are configured to generate a cycle value based on a target speed, based on a speed value associated with the motor at a particular time, and based on a current value associated with the motor at the particular time. The plurality of neurons is configured to be trained to generate the cycle value to minimize an error between the cycle value and a training cycle value for each training vector of a plurality of training vectors. The apparatus is configured to have generated the plurality of training vectors. The motor circuitry is configured to control, based on the cycle value, a set of switching elements to drive the motor.

Abstract: Embodiments of the present invention provide systems, methods, and computer storage media for facilitating generation of prediction models. In some embodiments, a predetermined number of parameter value sets is identified. Each parameter value set includes a plurality of parameter values that represent corresponding parameters within a time series model. The parameter values can be selected in accordance with stratified sampling to increase a likelihood of prediction accuracy. The parameter value sets are input into a time series model to generate a prediction value in accordance with observed time series data, and the parameter value set resulting in a least amount of prediction error can be selected and used to generate a time series prediction model (ARIMA, AR, MA, ARMA) with corresponding model parameters, such as p, q, and/or k, subsequently used to predict values.