A method for generating a federated learning model is provided. The method includes obtaining images; obtaining sorting results of the images; and generating a trained federated learning model by training a federated learning model to be trained according to the images and the sorting results. The federated learning model to be trained is obtained after pruning a federated learning model to be pruned, and a pruning rate of a convolution layer in the federated learning model to be pruned is automatically adjusted according to a model accuracy during the pruning.

There are provided systems and methods for a fallback artificial intelligence (AI) system for redundancy during system failover. A service provider may provide AI systems for automated decision-making, such as for risk analysis, marketing, and the like. An AI system may operate in a production computing environment in order to provide AI decision-making based on input data, for example, by providing an output decision. In order to provide redundancy to the production AI system, the service provider may train a fallback AI system using the input/output data pairs from the production AI system. This may utilize a deep neural network and a continual learning trainer. Thereafter, when a failover condition is detected for the production AI system, the service provider may switch from the production AI system to the fallback AI system, which may provide decision-making operations during failure of within the production computing environment.

The technical solution involves a board card including a storage component, an interface apparatus, a control component, and an artificial intelligence chip. The artificial intelligence chip is connected to the storage component, the control component, and the interface apparatus, respectively; the storage component is used to store data; the interface apparatus is used to implement data transfer between the artificial intelligence chip and an external device; and the control component is used to monitor a state of the artificial intelligence chip. The board card is used to perform an artificial intelligence operation.

A stuck-at fault mitigation method for resistive random access memory (ReRAM)-based deep learning accelerators, includes: confirming a distorted output value (Y0) due to a stuck-at fault (SAF) by using a correction data set in a pre-trained deep learning network, by means of ReRAM-based deep learning accelerator hardware; updating an average (μ) and a standard deviation (σ) of a batch normalization (BN) layer by using the distorted output value (Y0), by means of the ReRAM-based deep learning accelerator hardware; folding the batch normalization (BN) layer in which the average (μ) and the standard deviation (σ) are updated into a convolution layer or a fully-connected layer, by means of the ReRAM-based deep learning accelerator hardware; and deriving a normal output value (Y1) by using the deep learning network in which the batch normalization (BN) layer is folded, by means of the ReRAM-based deep learning accelerator hardware.

Novel and useful system and methods of several functional safety mechanisms for use in an artificial neural network (ANN) processor. The mechanisms can be deployed individually or in combination to provide a desired level of safety in neural networks. Multiple strategies are applied involving redundancy by design, redundancy through spatial mapping as well as self-tuning procedures that modify static (weights) and monitor dynamic (activations) behavior. The various mechanisms of the present invention address ANN system level safety in situ, as a system level strategy that is tightly coupled with the processor architecture. The NN processor incorporates several functional safety concepts which reduce its risk of failure that occurs during operation from going unnoticed. The mechanisms function to detect and promptly flag and report the occurrence of an error with some mechanisms capable of correction as well. The safety mechanisms cover data stream fault detection, software defined redundant allocation, cluster interlayer safety, cluster intralayer safety, layer control unit (LCU) instruction addressing, weights storage safety, and neural network intermediate results storage safety.

In one embodiment, a set of parameters representing a first state of an autonomous driving vehicle (ADV) to be simulated and a set of control commands to be issued at a first point in time. In response, a localization predictive model is applied to the set of parameters to determine a first position (e.g., x, y) of the ADV. A localization correction model is applied to the set of parameters to determine a set of localization correction factors (e.g., Δx, Δy). The correction factors may represent the errors between the predicted position of the ADV by the localization predictive model and the ground truth measured by sensors of the vehicle. Based on the first position of the ADV and the correction factors, a second position of the ADV is determined as the simulated position of the ADV.

Machine learning of model parameters for a neural network using a computing system is provided, that produces error-aware model parameters. An iterative process to converge on trained model parameters to be applied in the inference engine, includes applying a sequence of input training data sets to a neural network to produce inference results for the sequence using a set of model parameters in the neural network combined with factors based on a model of non-ideal characteristics of target memory to provide a training set of model parameters. An inference engine using the target memory technology to store the model parameters can have more stable results across a large number of engines.

The present disclosure relates to a neural network quantization parameter determination method and related products. A board card in the related products includes a memory device, an interface device, a control device, and an artificial intelligence chip, in which the artificial intelligence chip is connected with the memory device, the control device, and the interface device respectively. The memory device is configured to store data, and the interface device is configured to transmit data between the artificial intelligence chip and an external device. The control device is configured to monitor the state of the artificial intelligence chip. The board card can be used to perform an artificial intelligence computation.

In some implementations, the present disclosure relates to a method. The method includes obtaining a set of weights for a neural network comprising a plurality of nodes and a plurality of connections between the plurality of nodes. The method also includes identifying a first subset of weights and a second subset of weights based on the set of weights. The first subset of weights comprises weights that used by the neural network. The second subset of weights comprises weights that are prunable. The method further includes storing the first subset of weights in a first portion of a memory. A first error correction code is used for the first portion of the memory. The method further includes storing the second subset of weights in a second portion of the memory. A second error correction code is used for the second portion of the memory. The second error correction code is weaker than the first error correction code.

A device and method for machine learning using an artificial neural network. For a calculation hardware for the artificial neural network, a layer description is provided, which defines at least one part of a layer of the artificial neural network, the layer description defining a tensor for input values of at least one part of this layer, a tensor for weights of at least one part of this layer, and a tensor for output values of at least one part of this layer, in particular of its starting address. A message that includes a start address of the tensor for the input values, or of the tensor for the weighs, or of the tensor for the output values is sent by the calculation hardware for transfer of the input values, or the weights, or the output values, is sent by the calculation hardware.