Qualcomm Incorporated

Thomas Zheng

Yinyin Liu

Paul E Bender

Jason Frank Hunzinger

Victor Hokkiu Chan

Sreeram Kannan

Certain embodiments of the present disclosure relate to a technique for image reconstruction that employs cascaded over-complete dictionaries (i.e., collections of bases) for extracting features and building representations for images at different reconstruction levels. Each dictionary on a different reconstruction level can be learned and optimized for the purpose of capturing either generic or discriminative features. By finding sparse representations through the cascaded dictionaries, an image can be reconstructed and recognized.

Image recognition system based on cascaded over-complete dictionaries

Certain aspects of the present disclosure support a technique for strategic synaptic failure and learning in spiking neural networks. A synaptic weight for a synaptic connection between a pre-synaptic neuron and a post-synaptic neuron can be first determined (e.g., according to a learning rule). Then, one or more failures of the synaptic connection can be determined based on a set of characteristics of the synaptic connection. The one or more failures can be omitted from computation of a neuronal behavior of the post-synaptic neuron.

Method and apparatus for strategic synaptic failure and learning in spiking neural networks

Certain aspects of the present disclosure relate to methods and apparatus for inferring causal relationship between random processes using a temporal learning algorithm. The temporal learning algorithm determines structure of a causal graph with a set of nodes. Input to the nodes may be binary time series (e.g., random processes). The output of the temporal learning algorithm may be a labeled directed graph in which the direction of the connection between each two node indicates causal direction and the strength of connectivity between the nodes indicates intensity of the causal influence. The temporal learning algorithm may iteratively update strength of connections between nodes to track variations in real time.

Method and apparatus for inferring logical dependencies between random processes

Certain embodiments of the present disclosure relate to a method for face recognition that is occlusion tolerant and scale/shift invariant based on a combination of hierarchical maximization and adaptive representation technique.

Growing community of inventors

San Diego, CA, United States of America

Thomas Zheng