18510755. COUNTER BASED RESISTIVE PROCESSING UNIT FOR PROGRAMMABLE AND RECONFIGURABLE ARTIFICIAL-NEURAL-NETWORKS simplified abstract (INTERNATIONAL BUSINESS MACHINES CORPORATION)
Contents
- 1 COUNTER BASED RESISTIVE PROCESSING UNIT FOR PROGRAMMABLE AND RECONFIGURABLE ARTIFICIAL-NEURAL-NETWORKS
- 1.1 Organization Name
- 1.2 Inventor(s)
- 1.3 COUNTER BASED RESISTIVE PROCESSING UNIT FOR PROGRAMMABLE AND RECONFIGURABLE ARTIFICIAL-NEURAL-NETWORKS - A simplified explanation of the abstract
- 1.4 Simplified Explanation
- 1.5 Potential Applications
- 1.6 Problems Solved
- 1.7 Benefits
- 1.8 Potential Commercial Applications
- 1.9 Original Abstract Submitted
COUNTER BASED RESISTIVE PROCESSING UNIT FOR PROGRAMMABLE AND RECONFIGURABLE ARTIFICIAL-NEURAL-NETWORKS
Organization Name
INTERNATIONAL BUSINESS MACHINES CORPORATION
Inventor(s)
SIYURANGA Koswatta of Carmel NY (US)
YULONG Li of Westchester NY (US)
Paul M. Solomon of Westchester NY (US)
COUNTER BASED RESISTIVE PROCESSING UNIT FOR PROGRAMMABLE AND RECONFIGURABLE ARTIFICIAL-NEURAL-NETWORKS - A simplified explanation of the abstract
This abstract first appeared for US patent application 18510755 titled 'COUNTER BASED RESISTIVE PROCESSING UNIT FOR PROGRAMMABLE AND RECONFIGURABLE ARTIFICIAL-NEURAL-NETWORKS
Simplified Explanation
The abstract describes technical solutions for storing weight in a crosspoint device of a resistive processing unit (RPU) array. The system includes a crosspoint array where each node represents a connection between neurons of a neural network and stores a weight assigned to the node. The crosspoint device at each node includes a counter with multiple single bit counters, where the states of the counters represent the weight to be stored. Additionally, the crosspoint device includes a resistor device with multiple resistive circuits, each associated with a respective counter. The resistive circuits are activated or deactivated based on the state of the associated counter, adjusting the electrical conductance of the resistor device.
- Crosspoint array stores weights assigned to connections between neurons
- Crosspoint device at each node includes a counter with multiple single bit counters
- States of counters represent the weight to be stored
- Resistor device with multiple resistive circuits associated with counters
- Resistive circuits are activated or deactivated based on counter states to adjust electrical conductance
Potential Applications
This technology can be applied in:
- Neural network training
- Artificial intelligence systems
- Machine learning algorithms
Problems Solved
This technology addresses issues such as:
- Efficient weight storage in neural networks
- Improved processing capabilities
- Enhanced performance of resistive processing units
Benefits
The benefits of this technology include:
- Higher efficiency in storing weights
- Increased accuracy in neural network operations
- Enhanced performance in machine learning tasks
Potential Commercial Applications
- "Innovative Weight Storage Solutions for Neural Networks
Original Abstract Submitted
Technical solutions are described for storing weight in a crosspoint device of a resistive processing unit (RPU) array. An example system includes a crosspoint array, wherein each array node represents a connection between neurons of the neural network, and wherein each node stores a weight assigned to the node. The crosspoint array includes a crosspoint device at each node. The crosspoint device includes a counter that has multiple single bit counters, and states of the counters represent the weight to be stored at the crosspoint device. Further, the crosspoint device includes a resistor device that has multiple resistive circuits, and each resistive circuit is associated with a respective counter from the counters. The resistive circuits are activated or deactivated according to a state of the associated counter, and an electrical conductance of the resistor device is adjusted based at least in part on the resistive circuits that are activated.
