International Business Machines Corporation

Globalfoundries Inc.

Stmicroelectronics Gmbh

Samsung Electronics Co., Ltd.

Stmicroelectronics (crolles 2) Sas

Toshiba America Electronic Components, Inc.

Globalfoudries Inc.

Ahmet Serkan Ozcan

Christian Lavoie

Emre Alptekin

Kathryn Turner Schonenberg

Nicolas L Breil

Filippos Papadatos

Praneet Adusumilli

Zhen Zhang

Cung D Tran

Bin Yang

Paul Michael Solomon

Viraj Yashawant Sardesai

Shawn Peter Fetterolf

Kam-Leung Lee

Jian Yu

J Campbell Scott

Kangguo Cheng

Jack Oon Chu

Jeng-Bang Yau

Charles Edwin Cox

John Bruley

Hemanth Jagannathan

Sivananda K Kanakasabapathy

Chengwen Pei

Arvind Kumar

Zhengwen O Li

Nicolas Antoine

Oleg Gluschenkov

Haining S Yang

Deok-Kee Kim

Pritish Narayanan

Harald Huels

Winfried W Wilcke

Alexis Asseman

Tomasz Kornuta

Carl John Radens

Guy M Cohen

Xiao Hu Liu

Wayne Isami Imaino

Viorel C Ontalus

Lindsey H Hall

Donald Rene Wall

Dong-Ick Lee

Nicolas Degors

Michael Hatzistergos

Yiyi Wang

Devendra K Sadana

Tak H Ning

Keith Kwong Hon Wong

Katherine Lynn Saenger

Joel Pereira De Souza

Alberto Valdes Garcia

Yu Zhu

Stephen Mark Rossnagel

Jean-Olivier Plouchart

Roger Allan Quon

Jin-Ping Han

Anthony Gene Domenicucci

Paul Stephen McLaughlin

Mark V Raymond

Randolph F Knarr

Ming Cai

Daniel Waddington

Andrew John Kellock

Petar K Pepeljugoski

David F Hilscher

Nathaniel C Berliner

Stefan Zollner

Emmanuel Petitprez

Chaitanya Shivade

Masafumi Hamaguchi

Asa Frye

Asaf Tzadok

Deepta Rajan

Javier Ayala

Vincent Albouy

An imaging system is provided. A first imaging system captures initial sensor data in a form of visible domain data. A second imaging system captures subsequent sensor data in a form of second domain data, wherein the initial and subsequent sensor data are of different spectral domains. A controller subsystem detects at least one region of interest in real-time by applying a machine learning technique to the visible domain data, localizes at least one object of interest in the at least one region of interest to generate positional data for the at least one object of interest, and autonomously steers a point of focus of the second imaging system to a region of a scene including the object of interest to capture the second domain data responsive to the positional data.

Guided multi-spectral inspection

Feedback within hierarchical neural networks such as hierarchical temporal memories is provided. In various embodiments, a plurality of input streams is received at a first region of a hierarchical artificial neural network. The plurality of input streams is pooled in the first region. One or more recurrent features are identified in the pooled input streams in the first region. A temporal pattern is recognized in the one or more recurrent features in the first region. An indication of the temporal pattern is provided to at least a second region of the hierarchical neural network. A response to a further input is modulated based on feedback from the second region and at least one additional region.

Computational method for feedback in a hierarchical neural network

A method of forming a contact to a semiconductor device is provided that forms an alloy composed of nickel (Ni), platinum (Pt), aluminum (Al), titanium (Ti) and a semiconductor material. The methods may include forming a nickel and platinum semiconductor alloy at a base of a via. A titanium layer having an angstrom scale thickness is deposited in the via in contact with the nickel platinum semiconductor alloy. An aluminum containing fill is deposited atop the titanium layer. A forming gas anneal including an oxygen containing atmosphere is applied to the structure to provide a contact alloy comprising nickel, platinum, aluminum, titanium and a semiconductor element from the contact surface of the semiconductor device.

Low resistance contacts including intermetallic alloy of nickel, platinum, titanium, aluminum and type IV semiconductor elements

A system according to one embodiment includes a collection of computing nodes arranged in a mesh of N&#xd7;M&#xd7;Z topology, the nodes including computational hardware, wherein Z&#x3c;N and Z&#x3c;M, and wherein N and M are at least equal to 4; a collection of I/O connections interfaced with one of the sides of the mesh, said side having N&#xd7;M nodes, each of the I/O connections being tied to a unique one of the nodes in said side; and I/O cards that are tied to the I/O connections.

Configuring computing nodes in a three-dimensional mesh topology

A method for using a distributed memory device in a memory augmented neural network system includes receiving, by a controller, an input query to access data stored in the distributed memory device, the distributed memory device comprising a plurality of memory banks. The method further includes determining, by the controller, a memory bank selector that identifies a memory bank from the distributed memory device for memory access, wherein the memory bank selector is determined based on a type of workload associated with the input query. The method further includes computing, by the controller and by using content based access, a memory address in the identified memory bank. The method further includes generating, by the controller, an output in response to the input query by accessing the memory address.

Distributed memory-augmented neural network architecture

A computational method is disclosed for the simulation of a hierarchical artificial neural network (ANN), wherein a single correlator pools, during a single time-step, two or more consecutive feed-forward inputs from previously predicted and now active neurons of one or more lower levels.

Computational method for temporal pooling and correlation

According to embodiments of the present disclosure, methods of and computer program products for operating a plurality of classifiers are provided. A plurality of input entities are read, each input entity having an associated target label. The input entities are provided to a first classifier, and a category of each input entity is obtained therefrom. A feature map is determined for each input entity. Each feature map is provided to each of a set of classifiers, and an assigned label is obtained for each feature map from each of the set of classifiers. Each classifier is associated with one of the categories. For each classifier, the assigned label for each feature map is compared to the target labels to determine a plurality of gradients. The plurality of gradients are masked according to each category, yielding a masked set of gradients for each category. Each classifier is trained according its associated masked gradients.

Variational gradient flow

A computer-implemented method is provided for neuromorphic computing in a Dynamic Random Access Memory (DRAM). The method includes representing one or more neurons by memory rows in the DRAM. Each bit in any of the memory rows represents a single synapse. The method further includes responsive to activating a given one of the neurons, reinforcing an associated synaptic state of a corresponding one of the memory rows representing the given one of the neurons. The method also includes responsive to inhibiting the given one of the neurons, degrading the associated synaptic state of the corresponding one of the memory rows representing the given one of the neurons.

Neuromorphic computing in dynamic random access memory

A computer-implemented method is provided for use with a reconfigurable computational device having a collection of computing nodes arranged in a mesh of N&#xd7;M&#xd7;Z topology, the computing nodes including computational hardware, wherein Z&#x3c;N and Z&#x3c;M, and wherein N and M are at least equal to 4. The method includes using the computational device to perform computations characterized by (i) an initial system I/O bandwidth and (ii) an initial system node-to-node latency; reconfiguring the device into a mesh of N'&#xd7;M&#x2032;&#xd7;Z&#x2032; topology, wherein at least two of N, M, and Z values are different from their corresponding N&#x2032;, M&#x2032;, and Z&#x2032; values, and wherein N&#xd7;M&#xd7;Z is equal to N&#x2032;&#xd7;M&#x2032;&#xd7;Z&#x2032;; and using the device to perform computations characterized by (i) a modified system I/O bandwidth and (ii) a modified system node-to-node latency.

Embodiments of the present invention describe a hierarchical cortical emulation using a scratchpad memory device and a storage class memory device. The scratchpad memory device is partitioned into a first subset of memory locations and a second subset of memory locations. A processor from a neural network device is assigned a first memory portion from the first subset, a second memory portion from the second subset, and a third memory portion from the storage class memory device. Further the neural network device and a memory controller perform a compute cycle for a hierarchical level k, 1&#x2264;k&#x2264;n, n being total number of levels. A compute cycle includes performing, by the processor, computations from the level k using neuron data stored in the first memory portion, and in parallel, copying by the memory controller, the neuron data for a hierarchical level k+1 from the third memory portion to the second memory portion.

Virtualization in hierarchical cortical emulation frameworks

Homeostasis-maintaining binary neural networks such as hierarchical temporal memories are provided. In various embodiments, a region of an artificial neural network is initialized. The region comprises a plurality of neurons and has a permanence value associated with each potential synaptic connection between neurons. The initialization comprises connecting a subset of the potential synaptic connections by synapses. A plurality of time-ordered inputs to the region is received. Some of the plurality of neurons are thereby caused to fire upon receipt of each time-ordered input. Upon receipt of each time-ordered input, for each potential synaptic connection between neurons, the permanence value is adjusted according to a firing sequence of the plurality of neurons. Those potential synaptic connections having a permanence value above a predetermined permanence threshold are connected when the total number of connected synapses in the region does not exceed a predetermined connectivity threshold. Synapses whose associated pre-synaptic neuron or post-synaptic neuron have more than a predetermined number of connected synapses are disconnected.

Approaching homeostasis in a binary neural network

Self-adjusting thresholds for synaptic activity in neural networks are provided. In various embodiments, for each of a plurality of neurons within an artificial neural network, an overlap value is determined corresponding to active inputs connected to the neuron via synapses having non-zero synaptic weights. A count of those of the plurality of neurons whose overlap exceeds an activation threshold of the neural network is determined. The count is compared to a predetermined neuronal activity target. The activation threshold of the neural network is adjusted to approach the predetermined neuronal activity target.

Self-adjusting threshold for synaptic activity in neural networks

Growing community of inventors

Pleasantville, NY, United States of America

Ahmet Serkan Ozcan