Back to results
US 12,001,941 B2Granted
Wireless oscillator neural network
AssigneeIntel Corporation
InventorsDmitri E. Nikonov, Sasikanth Manipatruni, Ian A. Young
Life of this patent
Estimating the term…
Abstract
Embodiments may relate to a system to be used in an oscillating neural network (ONN). The system may include a control node and a plurality of nodes wirelessly communicatively coupled with a control node. A node of the plurality of nodes may be configured to identify an oscillation frequency of the node based on a weight W and an input X. The node may further be configured to transmit a wireless signal to the control node, wherein a frequency of the wireless signal oscillates based on the identified oscillation frequency. Other embodiments may be described or claimed.
Claims17
A neural network, comprising:
- a first sensor node comprising:
- a first sensor configured to generate a first sensor output based on detection of a first environmental condition, and
- a first processor configured to output a first signal that has a first oscillation frequency in a first locking range, the first signal generated based on the first sensor output;
- a second sensor node communicatively coupled with the first sensor node, the second sensor node comprising:
- a second sensor configured to generate a second sensor output based on detection of a second environmental condition, and
- a second processor configured to:
- receive the first signal from the first sensor node,
- output a second signal that has a second oscillation frequency in the first locking range, the second signal generated based on the second sensor output,
- determine a modified second oscillation frequency by synchronizing the second oscillation frequency in the first locking range with the first oscillation frequency in the first locking range, wherein synchronizing the second oscillation frequency with the first oscillation frequency comprises modifying the second oscillation frequency so that a difference between the modified second oscillation frequency and the first oscillation frequency is smaller than a difference between the second oscillation frequency and the first oscillation frequency, and
- output a modified second signal that has the modified second oscillation frequency;
- and
- a third sensor node configured to generate a third signal that has a third oscillation frequency in a second locking range, wherein the second locking range is different from the first locking range.
9 more elements
A sensor node comprising:
- a sensor configured to generate sensor data based on detection of an environmental condition;
- a processor coupled with the sensor, wherein the processor is configured to:
- receive a first signal from another sensor node communicatively coupled with the sensor node, the first signal having a first oscillation frequency in a first locking range,
- determine a second oscillation frequency based on the sensor data and a weight signal, and
- after determining that the second oscillation frequency is in the first locking range, determine a modified second oscillation frequency by synchronizing the second oscillation frequency with the first oscillation frequency, wherein synchronizing the second oscillation frequency with the first oscillation frequency comprises modifying the second oscillation frequency so that a difference between the modified second oscillation frequency and the first oscillation frequency is smaller than a difference between the second oscillation frequency and the first oscillation frequency;
- and
- a transmit module configured to:
- transmit a second signal having the second oscillation frequency, and
- transmit a modified second signal having the modified second oscillation frequency,
- wherein the sensor node and the another sensor node are in a neural network, the neural network further comprises a third sensor node configured to generate a third signal that has a third oscillation frequency in a second locking range, and the second locking range is different from the first locking range.
7 more elements
An oscillating neural network (ONN), comprising:
- a control node configured to process a plurality of wireless signals to generate an output of the ONN;
- and
- a plurality of sensor nodes configured to generate the plurality of wireless signals and to transmit the plurality of wireless signals to the control node, wherein the plurality of sensor node comprises:
- a first sensor node configured to:
- generate a first sensor output, and
- output a first signal that has a first oscillation frequency in a first locking range, the first signal generated based on the first sensor output,
- a second sensor node communicatively coupled with the first sensor node, the second sensor node configured to:
- generate a second sensor output,
- receive the first signal from the first sensor node,
- output a second signal that has a second oscillation frequency in the first locking range, the second signal generated based on the second sensor output,
- determine a modified second oscillation frequency by synchronizing the second oscillation frequency in the first locking range with the first oscillation frequency in the first locking range, wherein synchronizing the second oscillation frequency with the first oscillation frequency comprises modifying the second oscillation frequency so that a difference between the modified second oscillation frequency and the first oscillation frequency is smaller than a difference between the second oscillation frequency and the first oscillation frequency, and
- output a modified second signal that has the modified second oscillation frequency, and
- a third sensor node configured to generate a third signal that has a third oscillation frequency in a second locking range, wherein the second locking range is different from the first locking range.
10 more elements
Dependencies are inferred from the claim text.