Recognition, reidentification and security enhancements using autonomous machines
A mechanism is described for facilitating recognition, reidentification, and security in machine learning at autonomous machines. A method of embodiments, as described herein, includes facilitating a camera to detect one or more objects within a physical vicinity, the one or more objects including a person, and the physical vicinity including a house, where detecting includes capturing one or more images of one or more portions of a body of the person. The method may further include extracting body features based on the one or more portions of the body, comparing the extracted body features with feature vectors stored at a database, and building a classification model based on the extracted body features over a period of time to facilitate recognition or reidentification of the person independent of facial recognition of the person.
1. An apparatus comprising:
one or more processors including a graphics processor; and
a memory to store data including sensor data generated by one or more sensors;
wherein the one or more processors are to:
obtain the sensor data generated by the one or more sensors,
provide the sensor data as an input to a trained neural network model, and
generate a decision for an autonomous machine utilizing the neural network model based at least in part on the sensor data; and
wherein generating the decision includes providing protection in operation of the neural network model against an attack attempting to change an output of the neural network model, the protection including one or more integrity verification checks that are inserted into each of a plurality of layers of the neural network model, the one or more integrity verification checks to check integrity of each of the plurality of layers of the neural network to protect the neural network from attack.
2. The apparatus of claim 1 , wherein the protection further includes performing separate parallel executions of the neural network model and a software application associated with the one or more processors.
3. The apparatus of claim 2 , wherein the performance of the parallel execution includes the neural network being protected in a first processing unit of the one or more processors and the software application being quarantined in a second processing unit of the one or more processors.
4. The apparatus of claim 1 , wherein the protection further includes comparing an output of the neural network model with a pending decision of a decision-making entity.
5. The apparatus of claim 4 , wherein, based on the output of the neural network model, the pending decision is altered, suspended, or maintained.
6. The apparatus of claim 1 , wherein the autonomous machine includes an autonomous vehicle, and wherein the decision of the neural network model includes a decision for operation of the autonomous vehicle.
7. The apparatus of claim 1 , wherein the graphics processor is co-located with an application processor on a common semiconductor package.
8. The apparatus of claim 1 , wherein the one or more integrity verification checks include calculating a Cyclic Redundancy Check (CRC) and requiring a correct calculated CRC result for crossing each of the plurality of layers of the neural network model.
9. The apparatus of claim 1 , wherein the one or more integrity verification checks include requiring a security token to cross each of the plurality of layers of the of the neural network model.
10. A method comprising:
obtaining sensor data that is generated by one or more sensors for an autonomous machine;
providing the sensor data as an input to a trained neural network model in a system including one or more processors including a graphics processor; and
generating a decision for the autonomous machine utilizing the neural network model based at least in part on the sensor data;
wherein generating the decision includes providing protection in operation of the neural network model against an attack attempting to change an output of the neural network model, the protection including one or more integrity verification checks that are inserted into each of a plurality of layers of the neural network model, the one or more integrity verification checks to check integrity of each of the plurality of layers of the neural network to protect the neural network from attack.
11. The method of claim 10 , wherein the protection further includes performing separate parallel executions of the neural network model and a software application associated with the one or more processors.
12. The method of claim 11 , wherein the performance of the parallel execution includes the neural network being protected in a first processing unit of the one or more processors and the software application being quarantined in a second processing unit of the one or more processors.
13. The method of claim 10 , wherein the protection further includes comparing an output of the neural network model with a pending decision of a decision-making entity.
14. The method of claim 13 , wherein, based on the output of the neural network model, the pending decision is altered, suspended, or maintained.
15. At least one non-transitory machine-readable medium comprising instructions that when executed by a computing device, cause the computing device to perform operations comprising:
obtaining sensor data that is generated by one or more sensors for an autonomous machine;
providing the sensor data as an input to a trained neural network model in a system including one or more processors including a graphics processor; and
generating a decision for the autonomous machine utilizing the neural network model based at least in part on the sensor data;
wherein generating the decision includes providing protection in operation of the neural network model against an attack attempting to change an output of the neural network model, the protection including one or more integrity verification checks that are inserted into each of a plurality of layers of the neural network model, the one or more integrity verification checks to check integrity of each of the plurality of layers of the neural network to protect the neural network from attack.
16. The machine-readable medium of claim 15 , wherein the protection further includes performing separate parallel executions of the neural network model and a software application associated with the one or more processors, wherein the performance of the parallel execution includes the neural network being protected in a first processing unit of the one or more processors and the software application being quarantined in a second processing unit of the one or more processors.
17. The machine-readable medium of claim 15 , wherein the protection further includes comparing an output of the neural network model with a pending decision of a decision-making entity, wherein, based on the output of the neural network model, the pending decision is altered, suspended, or maintained.