A method and server for optimizing hyperparameter tuples for training production-grade artificial intelligence (AI) models. For each one of the AI models, AI model features are extracted and, for the one AI model, an initial distribution of n hyperparameter tuplesis created considering the extracted AI model features therefor. A loop is repeated, until metric parameters are satisfied, comprising: evaluating latency from training the one AI model for each of the n hyperparameters tuples; evaluating model uncertainty from training the one AI model for each of the n hyperparameters tuples; for each of the n hyperparameters tuples, computing a blended quality measurement from the evaluated latency and evaluated model uncertainty; replacing m hyperparameter tuples having the worst blended quality measurements with m newly generated hyperparameter tuples. The metric parameters include one or more of a threshold value on model uncertainty and blended quality measurement gain between successive loops.

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for selecting a neural network to perform a particular machine learning task while satisfying a set of constraints.

Methods and systems for reverse design of micro-nano structure based on a deep neural network. The method includes step 101 of acquiring initial data of a micro-nano structure according to the micro-nano structure to be reversely designed. The method also includes step 102 of inputting the initial data of the micro-nano structure into a trained optical parameter prediction model to obtain optical prediction parameters. The method further includes step 103 of evaluating the optical prediction parameters. The method also includes optimizing the initial data of the micro-nano structure, inputting the optimized data of the micro-nano structure into the trained optical parameter prediction model, and performing steps 102 and 103 again until the evaluation result of the optical prediction parameters obtained in a current iteration satisfies the preset condition. Through the method of the present application, the electromagnetic response calculation time of the reverse design is greatly shortened.

A super-network comprising a plurality of layers may be generated. Each layer may comprise cells with different structures. A predetermined number of cells from each layer may be selected. A plurality of cells may be generated based on selected cells using a local mutation model, wherein the local mutation model comprises a mutation window for removing redundant edges from each selected cell. Performance of the plurality of cells may be evaluated using a differentiable fitness scoring function. The operations of the generating a plurality of cells using the local mutation model, the evaluating performance of the plurality of cells using the differentiable fitness scoring function and the selecting the subset of cells based on the evaluation results may be iteratively performed until the super-network converges. A search space for each layer may be generated based on a predetermined top number of cells with largest fitness scores after the super-network converges.

The present disclosure relates to a spatially-variant model of a point spread function and its role in enhancing medical image resolution. For instance, a method of the present disclosure comprises receiving a first medical image having a first resolution, applying a neural network to the first medical image, the neural network including a first subset of layers and, subsequently, a second subset of layers, the first subset of layers of the neural network generating, from the first medical image, a second medical image having a second resolution and the second subset of layers of the neural network generating, from the second medical image, a third medical image having a third resolution, and outputting the third medical image, wherein the first resolution is lower than the second resolution and the second resolution is lower than the third resolution.

A method includes obtaining identifiers of entities and symbolic artificial intelligence (AI) models configured to produce outputs responsive to inputs based on events caused by at least one of the entities. At least some of the entities are associated with outputs of respective symbolic AI models and have respective scores corresponding to the respective outputs of the symbolic AI models. The method may include obtaining scenarios, where each scenario includes simulated inputs corresponding to one or more simulated events, and at least some scenarios include a plurality of simulated inputs. The method may also include determining a population of scores of a given entity among the entities, where respective members of the population of scores correspond to respective outputs of the plurality of symbolic AI models, and where the respective outputs correspond to respective scenarios among the scenarios and storing the population of scores in memory.

One or more multi-stage optimization iterations are performed with respect to a compression algorithm. A given iteration comprises a first stage in which hyper-parameters of a perceptual quality algorithm are tuned independently of the compression algorithm. A second stage of the iteration comprises tuning hyper-parameters of the compression algorithm using a set of perceptual quality scores generated by the tuned perceptual quality algorithm. The final stage of the iteration comprises performing a compression quality evaluation test on the tuned compression algorithm.

Methods, computer systems, and apparatus, including computer programs encoded on computer storage media, for generating and searching reinforcement learning algorithms. In some implementations, a computer-implemented system generates a sequence of candidate reinforcement learning algorithms. Each candidate reinforcement learning algorithm in the sequence is configured to receive an input environment state characterizing a state of an environment and to generate an output that specifies an action to be performed by an agent interacting with the environment. For each candidate reinforcement learning algorithm in the sequence, the system performs a performance evaluation for a set of a plurality of training environments. For each training environment, the system adjusts a set of environment-specific parameters of the candidate reinforcement learning algorithm by performing training of the candidate reinforcement learning algorithm to control a corresponding agent in the training environment. The system generates an environment-specific performance metric for the candidate reinforcement learning algorithm that measures a performance of the candidate reinforcement learning algorithm in controlling the corresponding agent in the training environment as a result of the training. After performing training in the set of training environments, the system generates a summary performance metric for the candidate reinforcement learning algorithm by combining the environment-specific performance metrics generated for the set of training environments. After evaluating each of the candidate reinforcement learning algorithms in the sequence, the system selects one or more output reinforcement learning algorithms from the sequence based on the summary performance metrics of the candidate reinforcement learning algorithms.

The disclosure discloses a method for detecting an intrusion in parallel based on an unbalanced data Deep Belief Network, which reads an unbalanced data set DS; under-samples the unbalanced data set using the improved NCR algorithm to reduce the ratio of the majority type samples and make the data distribution of the data set balanced; the improved differential evolution algorithm is used on the distributed memory computing platform Spark to optimize the parameters of the deep belief network model to obtain the optimal model parameters; extract the feature of data of the data set, and then classify the intrusion detection by the weighted nuclear extreme learning machine, and finally train multiple weighted nuclear extreme learning machines of different structures in parallel by multithreading as the base classifier, and establish a multi-classifier intrusion detection model based on adaptive weighted voting for detecting the intrusion in parallel.

A method for searching for an output machine learning (ML) algorithm to perform an ML task is described. The method includes: receiving a set of training examples and a set of validation examples, and generating a sequence of candidate ML algorithms to perform the task. For each candidate ML algorithm in the sequence, the method includes: setting up one or more training parameters for the candidate ML algorithm by executing a respective candidate setup function, training the candidate ML algorithm by processing the set of training examples using a respective candidate predict function and a respective candidate learn function, and evaluating a performance of the trained candidate ML algorithm by executing the respective candidate predict function on the set of validation examples to determine a performance metric. The method includes selecting a trained candidate ML algorithm with the best performance metric as the output ML algorithm for the task.