Preferred Networks, Inc.

Seiya Tokui

Ryosuke Okuta

Kenta Oono

Yuya Unno

Daisuke Okanohara

Toru Nishikawa

Justin B Clayton

Shohei Hido

Nobuyuki Ota

Nobuyuki Kubota

A compiler, for generating machine code to be executed in a chip including a plurality of distributed memories connected by a tree structure topology, includes at least one memory and at least one processor. The at least one processor is configured to associate each element of a tensor to be processed with an address in the plurality of memories included in the chip, based on a stride and a number of divisions in a predetermined hierarchy of the tree structure with respect to the tensor to be processed.

Compiler, generation method, chip, and execution method

There is provided an information processing device which efficiently executes machine learning. The information processing device according to one embodiment includes: an obtaining unit which obtains a source code including a code which defines Forward processing of each layer constituting a neural network; a storage unit which stores an association relationship between each Forward processing and Backward processing associated with each Forward processing; and an executing unit which successively executes each code included in the source code, and which calculates an output value of the Forward processing defined by the code based on an input value at a time of execution of each code, and generates a reference structure for Backward processing in a layer associated with the code based on the association relationship stored in the storage unit.

Information processing device and information processing method

Machine learning with model filtering and model mixing for edge devices in a heterogeneous environment is disclosed. In an example embodiment, an edge device includes a communication module, a data collection device, a memory, a machine learning module, and a model mixing module. The edge device analyzes collected data with a model for a first task, outputs a result, and updates the model to create a local model. The edge device communicates with other edge devices in a heterogeneous group, transmits a request for local models to the heterogeneous group, and receives local models from the heterogeneous group. The edge device filters the local models by structure metadata, including second local models, which relate to a second task. The edge device performs a mix operation of the second local models to generate a mixed model which relates to the second task, and transmits the mixed model to the heterogeneous group.

Machine learning with model filtering and model mixing for edge devices in a heterogeneous environment

A machine learning heterogeneous edge device, method, and system are disclosed. In an example embodiment, an edge device includes a communication module, a data collection device, a memory, a machine learning module, a group determination module, and a leader election module. The edge device analyzes collected data with a model, outputs a result, and updates the model to create a local model. The edge device communicates with other edge devices in a heterogeneous group. The edge device determines group membership and determines a leader edge device. The edge device receives a request for the local model, transmits the local model to the leader edge device, receives a mixed model created by the leader edge device performing a mix operation of the local model and a different local model, and replaces the local model with the mixed model.

Growing community of inventors

Chiba, Japan

Seiya Tokui