Methods, apparatus and computer-readable media related to data transmission in a neural network

Patent metadata

Field	Value
Publication number	US-2025013858-A1
Application number	US-202118706456-A
Country	US
Kind code	A1
Filing date	Nov 3, 2021
Priority date	Nov 3, 2021
Publication date	Jan 9, 2025
Grant date	—

Abstract

A method, performed by a transmitting node of a neural network, is provided for congestion level control. The method includes: sending, to a receiving node of the neural network, a plurality of sequences of data, the sequences of data being temporally encoded according to an encoding configuration; sending, to the receiving node, an indication of the encoding configuration, enabling the sequences of data to be decoded; receiving, from the receiving node, feedback including an indication of an error experienced in decoding the sequences of data; and adapting the encoding configuration based on the feedback.

First claim

1 . A method performed by a transmitting node of a neural network for congestion level control, the method comprising:
sending, to a receiving node of the neural network, a plurality of sequences of data, the sequences of data being temporally encoded according to an encoding configuration; sending, to the receiving node, an indication of the encoding configuration, enabling the sequences of data to be decoded; receiving, from the receiving node, feedback comprising an indication of an error experienced in decoding the sequences of data; and adapting the encoding configuration based on the feedback.

2 . The method of claim 1 , wherein the neural network comprises a spiking neural network, and wherein the sequences of data comprise sequences of one or more spikes.

3 . The method of claim 1 , wherein adapting the encoding configuration comprises adapting a rate at which information is encoded temporally for transmission.

4 . The method of claim 1 , wherein the encoding configuration comprises an inhibition time window during which the transmitting node is prohibited from transmitting one or more of the sequences of data.

5 . The method of claim 4 , wherein adapting the encoding configuration comprises adapting one or more of: a duration of the inhibition time window; and a periodicity of the inhibition time window.

6 . The method of claim 4 , wherein the indication of the encoding configuration comprises an indication of the inhibition time window.

7 . The method of claim 1 , wherein adapting the encoding configuration comprises:
adapting a rate at which data is transmitted by the transmitting node, such that an absolute rate at which data is transmitted by the transmitting node is changed for two or more sequences of data, but a relative rate between the two or more sequences of data is unchanged.

8 . The method of claim 7 , wherein adapting the rate at which data is transmitted by the transmitting node comprises:
transcoding the sequences of data according to an adaptive tuning curve, wherein the adaptive tuning curve adapts the average rate at which data is transmitted by the transmitting node.

9 .- 11 . (canceled)

12 . The method according to claim 1 , wherein adapting the encoding configuration comprises:
adapting a maximum rate at which data can be transmitted by the transmitting node.

13 . The method of claim 1 , wherein adapting the encoding configuration based on the feedback comprises applying a first encoding configuration when the error is a first level, and applying a second encoding configuration when the error is a second level, wherein the first level is greater than the second level, and wherein the first encoding configuration has a lower average data rate than the second encoding configuration.

14 . The method of claim 1 , wherein the sequences of data are sent to the receiving node using one or more of: radio transmission; optical transmission; free space visible light transmission; and electrical transmission.

15 . A method performed by a receiving node of a neural network for congestion level control, the method comprising:
receiving, from a transmitting node of a neural network and, a plurality of sequences of data, the sequences of data being temporally encoded according to an encoding configuration; receiving, from the transmitting node, an indication of the encoding configuration; decoding the sequences of data using the indication of the encoding configuration to obtain decoded data values for the sequences of data; and ending, to the transmitting node, feedback comprising an indication of an error associated with the decoding of the sequences of data.

16 . The method of claim 15 , wherein the neural network comprises a spiking neural network, and wherein the sequences of data comprise sequences of one or more spikes.

17 . The method of claim 15 , wherein the encoding configuration is based on the feedback, and wherein a first encoding configuration is applied when the error is a first level, and a second encoding configuration is applied when the error is a second level, wherein the first level is greater than the second level, and wherein the first encoding configuration has a lower average data rate than the second encoding configuration.

18 . The method of claim 15 , further comprising estimating the error, wherein the error comprises a difference between the sequences of data as received by the receiving node, and the sequences of data as transmitted by the transmitting node.

19 . The method of claim 18 , wherein the sequences of data comprise scalar values, and wherein estimating the error comprises averaging the scalar values over a time period.

20 . The method of claim 18 , wherein the sequences of data comprise sequences of data symbols selected from a predefined set of data symbols, and wherein estimating the error comprises measuring a distance between a data symbol in a received sequence of data and a nearest data symbol in the predefined set of data symbols.

21 . The method of claim 15 , wherein the sequences of data are received from the transmitting node using one or more of: radio transmission; optical transmission; free space visible light transmission; and electrical transmission.

22 . (canceled)

23 . A transmitting node for a neural network, the transmitting node comprising processing circuitry and a non-transitory computer-readable medium storing instructions which, when executed by the processing circuitry, cause the transmitting node to:
send, to a receiving node of the neural network, a plurality of sequences of data, the sequences of data being temporally encoded according to an encoding configuration; send, to the receiving node, an indication of the encoding configuration, enabling the sequences of data to be decoded; receive, from the receiving node, feedback comprising an indication of an error experienced in decoding the sequences of data; and adapt the encoding configuration based on the feedback.

24 .- 30 . (canceled)

31 . A receiving node for a neural network, the receiving node comprising processing circuitry and a non-transitory computer-readable medium storing instructions which, when executed by the processing circuitry, cause the receiving node to:
receive, from a transmitting node of a neural network and, a plurality of sequences of data, the sequences of data being temporally encoded according to an encoding configuration; receive, from the transmitting node, an indication of the encoding configuration; decode the sequences of data using the indication of the encoding configuration to obtain decoded data values for the sequences of data; and send, to the transmitting node, feedback comprising an indication of an error associated with the decoding of the sequences of data.

32 .- 38 . (canceled)

Assignees

• Ericsson Telefon Ab L M

Inventors

• Veres András
• Rácz András
• Borsos Tamas
• Parkvall Stefan
• Baldemair Robert

Classifications

• G06N3/063

  using electronic means · CPC title

• G06N3/049 Primary

  Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs · CPC title

• H04L1/0002 Primary

  by adapting the transmission rate · CPC title