US 2024/0404269 A1 · Inferencing and Learning Based on Sensorimotor Input Data

Abstract

One or more multi-layer systems are used to perform inference. A multi-layer system may correspond to a node that receives a set of sensory input data for hierarchical processing, and may be grouped to perform processing for sensory input data. Inference systems at lower layers of a multi-layer system pass representation of objects to inference systems at higher layers. Each inference system can perform inference and form their own versions of representations of objects, regardless of the level and layer of the inference systems. The set of candidate objects for each inference system is updated to those consistent with feature-location representations for the sensors as well as object representations at lower layers. The set of candidate objects is also updated to those consistent with candidate objects from other inference systems, such as inference systems at other layers of the hierarchy or inference systems included in other multi-layer systems.

Claims18

Dependent claims

A computer-implemented method of learning for inference, the method comprising:

processing, by a first input processor of a first inference system, first inputs indicating detection of a first feature from a first sensor associated with the first inference system;
selecting, at a first output processor of the first inference system coupled to the first input processor, a set of output elements for activation, the activated set of output elements representing an object;
associating a subset of the activated output elements in the first output processor with each other by updating first lateral connections between the subset of the activated output elements;
generating, by the first input processor, a first input representation by selecting a set of input elements for activation, the first input representation representing a first set of pairs indicating detection of the first feature of the object at a first location associated with the first feature;
and
associating a subset of the activated input elements in the first input processor with one or more of the activated output elements in the first output processor by updating feedforward connections.

3 more elements

A non-transitory computer readable storage medium comprising instructions thereon, the instructions when executed by one or more processors cause the one or more processors to:

process, by a first input processor of a first inference system, first inputs indicating detection of a first feature from a first sensor associated with the first inference system;
select, at a first output processor of the first inference system coupled to the first input processor, a set of output elements for activation, the activated set of output elements representing an object;
associate a subset of the activated output elements in the first output processor with each other by updating first lateral connections between the subset of the activated output elements;
generate, by the first input processor, a first input representation by selecting a set of input elements for activation, the first input representation representing a first set of pairs indicating detection of the first feature of the object at a first location associated with the first feature;
and
associate a subset of the activated input elements in the first input processor with one or more of the activated output elements in the first output processor by updating feedforward connections.

3 more elements

Dependencies are inferred from the claim text.

Inferencing and Learning Based on Sensorimotor Input Data

Life of this patent

Inferencing and Learning Based on Sensorimotor Input Data

Life of this patent

Abstract

Claims18