Abstract: This disclosure relates to testing of integrated artificial intelligence (AI) circuit with embedded memory to improve effective chip yield and to mapping addressable memory segments of the embedded memory to multilayer AI networks at the network level, layer level, parameter level, and bit level based on bit error rate (BER) of the addressable memory segments. The disclosed methods and systems allows for deployment of one or more multilayer AI networks in an AI circuit with sufficient model accuracy even when the embedded memory has an overall BER higher than a preferred overall threshold.

Abstract: This disclosure relates to testing of integrated artificial intelligence (AI) circuit with embedded memory to improve effective chip yield and to mapping addressable memory segments of the embedded memory to multilayer AI networks at the network level, layer level, parameter level, and bit level based on bit error rate (BER) of the addressable memory segments. The disclosed methods and systems allows for deployment of one or more multilayer AI networks in an AI circuit with sufficient model accuracy even when the embedded memory has an overall BER higher than a preferred overall threshold.

Abstract: This disclosure relates to artificial intelligence (AI) circuits with embedded memory for storing trained AI model parameters. The embedded memory cell structure, device profile, and/or fabrication process are designed to generate binary data access asymmetry and error rate asymmetry between writing binary zeros and binary ones that are adapted to and compatible with a binary data asymmetry of the trained model parameters and/or a bit-inversion tolerance asymmetry of the AI model between binary zeros and ones. The disclosed method and system improves predictive accuracy and memory error tolerance without significantly reducing an overall memory error rate and without relying on memory cell redundancy and error correction codes.

Abstract: This disclosure relates to a self-contained and self-sufficient edge device capable of performing processing data sets using a convolutional neural network model without relying on any backend servers. In particularly, the edge device may include non-volatile memory cells for storing a full set of trained model parameters from the convolutional neural network model. The non-volatile memory cells may be based on magnetic random access memory cells and may be embedded on the same semiconductor substrate with a convolutional neural network logic circuit dedicated to parallel forward propagation calculation.

Abstract: An apparatus according to one embodiment includes a read sensor. The read sensor has an antiferromagnetic layer (AFM), a first antiparallel magnetic layer (AP1 ) positioned above the AFM layer in a first direction oriented along a media-facing surface and perpendicular to a track width direction, a non-magnetic layer positioned above the AP1 in the first direction, a second antiparallel magnetic layer (AP2) positioned above the non-magnetic layer in the first direction, a harrier layer positioned above the AP2 in the first direction, and a free layer positioned above the barrier layer in the first direction. A soft bias layer is positioned behind at least a portion of the free layer in an element height direction normal to the media-facing surface, the soft bias layer including a soft magnetic material configured to compensate for a magnetic coupling of the free layer with the AP2.