Neuromorphic system for authorized user detection

Patent metadata

Field	Value
Publication number	US-10902115-B2
Application number	US-201916380687-A
Country	US
Kind code	B2
Filing date	Apr 10, 2019
Priority date	Oct 28, 2015
Publication date	Jan 26, 2021
Grant date	Jan 26, 2021

Abstract

Described is neuromorphic system for authorized user detection. The system includes a client device comprising a plurality of sensor types providing streaming sensor data and one or more processors. The one or more processors include an input processing component and an output processing component. A neuromorphic electronic component is embedded in or on the client device for continuously monitoring the streaming sensor data and generating out-spikes based on the streaming sensor data. Further, the output processing component classifies the streaming sensor data based on the out-spikes to detect an anomalous signal and classify the anomalous signal.

First claim

What is claimed is:

1. A neuromorphic system for authorized user detection, comprising:
a neuromorphic electronic component for embedding in or attached to a client device, the neuromorphic electronic component having a neuromorphic chip operable for continuously monitoring streaming sensor data from a client device and generating out-spikes based on the streaming sensor data;
a client device comprising a plurality of sensor types for providing the streaming sensor data, an input processing component, and an output processing component, and wherein the output processing component classifies the streaming sensor data based on the out-spikes to detect a user-transition; and
wherein the input processing component is configured to further perform operations of:
normalizing the streaming sensor data from the plurality of sensor types as a normalized time series;
grouping the normalized time series from the plurality of sensor types into a single scalar quantity;
collecting a grouping of samples of the single scalar quantity into a queue;
transforming the queue into discrete one-dimensional frequency domain data;
modifying the one-dimensional frequency domain data with a window function to reduce spectral leakage and generate frequency domain modified data;
filtering the frequency domain modified data to obtain scaled windowed frequency bins;
scaling both the normalized time series and scaled windowed frequency to generate in-rates;
mapping the in-rates to a distribution function; and
generating in-spikes based on the mapped in-rates.

2. The neuromorphic system as set forth in claim 1 , wherein the neuromorphic electronics component generates the out-spikes based on the in-spikes.

3. The neuromorphic system as set forth in claim 2 , wherein the neuromorphic electronics component generates the out-spikes based on the in-spikes using a randomly connected excitatory-inhibitory spiking network.

4. A neuromorphic system for authorized user detection, comprising:
a neuromorphic electronic component for embedding in or attached to a client device, the neuromorphic electronic component having a neuromorphic chip operable for continuously monitoring streaming sensor data from a client device and generating out-spikes based on the streaming sensor data;
a client device comprising a plurality of sensor types for providing the streaming sensor data, an input processing component, and an output processing component, and wherein the output processing component classifies the streaming sensor data based on the out-spikes to detect a user-transition; and
wherein the output processing component further performs operations of:
smoothing the out-spikes into rates;
applying the rates to a linear classifier to generate readouts;
filtering the readouts; and
classifying streaming sensor data as an anomalous signal based on the readouts.

5. The neuromorphic system as set forth in claim 1 , wherein upon classification of an anomalous signal, the output processing component further performs at least one operation of:
locking or unlocking access to the client device;
starting a new processing task;
executing a new logic branch of executable code;
transmitting information associated with the anomalous signal;
saving information associated with the anomalous signal to memory storage.

6. A neuromorphic method for authorized user detection, comprising acts of:
continuously monitoring streaming sensor data from a client device using a neuromorphic electronic component with a neuromorphic chip; and
generating out-spikes based on the streaming sensor data with the neuromorphic electronic component with the neuromorphic chip;
classifying the streaming sensor data based on the out-spikes to detect a user-transition using the client device, the client device having a plurality of sensor types for providing the streaming sensor data, an input processing component, and an output processing component, wherein the output processing component classifies the streaming sensor data; and
further performing, by the input processing component, acts of:
normalizing the streaming sensor data from the plurality of sensor types as a normalized time series;
grouping the normalized time series from the plurality of sensor types into a single scalar quantity;
collecting a grouping of samples of the single scalar quantity into a queue;
transforming the queue into discrete one-dimensional frequency domain data;
modifying the one-dimensional frequency domain data with a window function to reduce spectral leakage and generate frequency domain modified data;
filtering the frequency domain modified data to obtain scaled windowed frequency bins;
scaling both the normalized time series and scaled windowed frequency to generate in-rates;
mapping the in-rates to a distribution function; and
generating in-spikes based on the mapped in-rates.

7. The neuromorphic method as set forth in claim 6 , further comprising an act of generating, by the neuromorphic electronics component, the out-spikes based on the in-spikes.

8. The neuromorphic method as set forth in claim 7 , wherein the neuromorphic electronics component generates the out-spikes based on the in-spikes using a randomly connected excitatory-inhibitory spiking network.

9. The neuromorphic method as set forth in claim 6 , further performing, by the output processing component, acts of:
smoothing the out-spikes into rates;
applying the rates to a linear classifier to generate readouts;
filtering the readouts; and
classifying streaming sensor data as an anomalous signal based on the readouts.

10. The neuromorphic method as set forth in claim 6 , wherein upon classification of an anomalous signal, further performing at least one act of:
locking access to the client device;
starting a new processing task;
executing a new logic branch of executable code;
transmitting information associated with the anomalous signal; and
saving information associated with the anomalous signal to memory storage.

11. A computer program product for authorized user detection, the computer program product comprising:
a non-transitory computer-readable medium having executable instructions encoded thereon, such that upon execution of the instructions by one or more processors, the one or more processors perform operations of:
receiving out-spikes from a neuromorphic electronic component with a neuromorphic chip, the neuromorphic electronic component continuously monitoring streaming sensor data from a client device and generating out-spikes based on the streaming sensor data;
classifying, using the client device, the streaming sensor data based on the out-spikes to detect a user-transition, the client device having a plurality of sensor types for providing the streaming sensor data, an input processing component, and an output processing component, wherein the output processing component classifies the streaming sensor data;

further comprising instructions for causing the input processing component to perform operations of:
normalizing the streaming sensor data from the plurality of sensor types as a normalized time series;
grouping the normalized time series from the plurality of sensor types into a single scalar quantity;
collecting a grouping of samples of the single scalar quantity into a queue;
transforming the queue into discrete one-dimensional frequency domain data;
modifying the one-dimensional frequency domain data with a window function to reduce spectral leakage and generate frequency domain modified data;
filtering the frequency domain modified data to obtain scaled windowed frequency bins;
scaling both the normalized time series and scaled wi

Assignees

• Hrl Lab Llc
• Boeing Co

Inventors

• Patrick Richard J
• Stepp Nigel D
• De Sapio Vincent
• Cruz-Albrecht Jose
• Haley Jr John Richard
• Trostel Thomas M

Classifications

• G06N3/047

  Probabilistic or stochastic networks · CPC title

• G06N7/01

  Probabilistic graphical models, e.g. probabilistic networks · CPC title

• G06N3/0495

  Quantised networks; Sparse networks; Compressed networks · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06F21/554

  involving event detection and direct action · CPC title