Controller systems and methods of limiting the operation of neural networks to be within one or more conditions
Systems and methods for automatically self-correcting or correcting in real-time one or more neural networks after detecting a triggering event, or breaching boundary conditions are provided. Such a triggering event may indicate incorrect output signal or data being generated by the one or more neural networks. In particular, machine controllers of the invention limit the operations of neural networks to be within boundary conditions. Autonomous machines of the invention can be self-corrected after a breach of a boundary condition is detected. Autonomous land vehicles of the invention are capable of determining the timing of automatic transition to the manual control from automated driving mode. The controller of the invention filters and saves input-output data sets that fall within boundary conditions for later training of neural networks. The controllers of the invention include security architectures to prevent damages from virus attacks or system malfunctions.
1. An apparatus having a plurality of components, comprising:
an input device constructed to generate input data;
a first neural network with at least one hidden layer structured and trained to generate an output by processing input data received from the input device;
a first processor configured to receive the output from the first neural network and coupled to one of the components of the apparatus, the first processor further configured to control an operation of the one of the components of the apparatus using the output from the first neural network;
a second neural network, with at least one hidden layer, configured to receive the output from the first neural network, the second neural network structured and trained to generate a control output from the output of the first neural network; and
a second processor configured to:
receive the control output from the second neural network, and
control an operation of the first neural network using the control output from the second neural network, wherein said control an operation of the first neural network includes at least one of terminating the operation of the first neural network or re-training the first neural network.
2. The apparatus of claim 1 , wherein the second neural network is further structured to determine whether the output of the first neural network corresponds to a predetermined condition.
3. The apparatus of claim 2 , wherein the second processor is configured to terminate operation of the first neural network when the second neural network determines the output of the first neural network corresponds to the predetermined condition.
4. The apparatus of claim 2 , wherein the second processor is configured to re-train the first neural network when the output of the first neural network corresponds to the predetermined condition, wherein re-training of the first neural network includes retrieving a stored training data set, placing the first neural network in a retraining mode, retraining the first neural network using the retrieved training data set, and placing the first neural network in an operational mode.
5. The apparatus of claim 4 , wherein the second processor is further configured to re-train the first neural network using the input data generated by the input device.
6. The apparatus of claim 1 , further comprising:
a third neural network with at least one hidden layer, the third neural network configured to provide a second output to the second neural network, structured and trained to generate the second output by processing the input data,
wherein the second neural network is further trained to compare the output from the first neural network and the second output from the third neural network, and detect a difference between the output from the first neural network and the second output from the third neural network, and
wherein the second processor is further configured to perform a control action if the detected difference corresponds to a predetermined level, wherein said control an operation of the first neural network includes at least one of terminating the operation of the first neural network or re-training the first neural network.
7. The apparatus of claim 6 , wherein the second processor is configured to terminate the operation of the first neural network if the detected difference is above the predetermined level.
8. The apparatus of claim 6 , wherein the second processor is configured to re-train the first neural network if the detected difference is above the predetermined level.
9. The apparatus of claim 1 , wherein the apparatus comprises an autonomous land vehicle and the step of using the obtained output further includes the step of generating a message to control the autonomous land vehicle.
10. A computer-implemented method of controlling an apparatus, the method comprising:
processing input data using a first neural network with at least one hidden layer structured and trained to generate an output by inferencing on the input data;
controlling an operation of an aspect of the apparatus using the output from the first neural network;
generating a control output using a second neural network with at least one hidden layer structured to and trained to generate the control output by inferencing on the output generated by the first neural network; and
controlling an operation of the first neural network using the control output from the second neural network, wherein controlling an operation of the first neural network includes at least one of terminating the operation of the first neural network, re-training the first neural network,
wherein the method is performed by one or more computer hardware processors configured to execute computer-executable instructions on a non-transitory computer storage medium.
11. The computer-implemented method of claim 10 , further comprising:
determining whether the output of the first neural network corresponds to a predetermined condition; and
in response to determining that the output corresponds to the predetermined condition, performing an action affecting the operation of the first neural network.
12. The computer-implemented method of claim 11 , further comprising terminating the first neural network when the output is determined to correspond to the predetermined condition.
13. The computer-implemented method of claim 11 , further comprising re-training the first neural network when the output is determined to correspond to the predetermined condition.
14. The computer-implemented method of claim 11 , further comprising generating a human recognizable notification in response to determining the output corresponds to the predetermined condition.
15. The computer-implemented method of claim 11 , wherein the the predetermined condition is a condition that indicates the apparatus could be damaged and performing the action prevents damage to the apparatus.
16. The computer-implemented method of claim 10 , wherein the apparatus includes a human speech generator comprising a loudspeaker, and controlling the operation using the output includes generating human speech parts to be played on the loudspeaker.
17. The computer-implemented method of claim 10 , wherein the apparatus includes an autonomous land vehicle, and controlling the operation using the output includes generating a signal to control an autonomous land vehicle based at least in part on the output.
18. The computer-implemented method of claim 10 , wherein the apparatus comprises an image generator.
19. An apparatus having a plurality of components, the apparatus comprising:
an input device constructed to generate input data;
a first neural network with at least one hidden layer structured and trained to generate an output by processing input data from the input device;
a first processor configured to receive the output from the first neural network and coupled to one of the components of the apparatus, the first processor further configured to control an operation of the one of the components of the apparatus using the output from the first neural network;
a second neural network with at least one hidden layer configured to receive the output from the first neural network, the second neural network structured and trained to generate a control output and to determine whether the output corresponds to a predetermined condition; and
a second processor configured to receive the control output from the second neural network, the second processor configured to control an operation of the first neural network using the control output from the second neural network when the output of the first neural network exceeds the predetermined condition, wherein controlling an operation of the first neural network includes at least one of terminating the operation of the first neural network or re-training the first neural network.
20. The apparatus of claim 19 , wherein the second processor is further configured to terminate the first neural network when the second neural network determines the output from the first neural network corresponds to the predetermined condition.
21. The apparatus of claim 6 , wherein the third neural network is trained identical to the first neural network.