# US 2018/0322365 A1 · On the fly deep learning in machine learning for autonomous machines

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2018/0322365 A1
- Document ID: US20180322365A1
- Status: Published application
- Kind: A1
- Application number: 15/659,818
- Filed: 2017-07-26
- Published: 2018-11-08
- Earliest claimed: 2017-05-05
- Assignees: Intel Corporation
- Applicants: Intel Corporation
- Inventors: RAANAN YONATAN YEHEZKEL ROHEKAR
- Web page: https://patentgrep.com/patent/US20180322365A1

## CPC

- G06N 3/063: using electronic means
- G06F 9/46: Multiprogramming arrangements
- G06F 18/214: Generating training patterns; Bootstrap methods, e.g. bagging or boosting
- G06F 18/2148: {characterised by the process organisation or structure, e.g. boosting cascade}
- G06F 18/2411: based on the proximity to a decision surface, e.g. support vector machines
- G06F 18/2413: based on distances to training or reference patterns
- G06N 3/044: Recurrent networks, e.g. Hopfield networks
- G06N 3/045: Combinations of networks
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/08: Learning methods
- G06N 3/084: Backpropagation, e.g. using gradient descent
- G06N 3/0895: Weakly supervised learning, e.g. semi-supervised or self-supervised learning
- G06N 3/09: Supervised learning
- G06V 10/454: {Integrating the filters into a hierarchical structure, e.g. convolutional neural networks [CNN]}
- G06V 10/764: using classification, e.g. of video objects
- G06V 10/774: Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
- G06V 10/82: using neural networks
- G06V 10/95: {structured as a network, e.g. client-server architectures}
- G06V 10/955: {using specific electronic processors}
- G06V 20/00: Scenes; Scene-specific elements (control of digital cameras H04N23/60)
- G06F 9/505: {considering the load}
- G06V 40/174: {Facial expression recognition}
- G06V 2201/06: Recognition of objects for industrial automation
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING

## Abstract

A mechanism is described for facilitating on-the-fly deep learning in machine learning for autonomous machines. A method of embodiments, as described herein, includes detecting an output associated with a first deep network serving as a user-independent model associated with learning of one or more neural networks at a computing device having a processor coupled to memory. The method may further include automatically generating training data for a second deep network serving as a user-dependent model, where the training data is generated based on the output. The method may further include merging the user-independent model with the user-dependent model into a single joint model.

## Claims

### Claim 1

An apparatus comprising:
detection/observation logic, as facilitated by or at least partially incorporated into a processor, to detect an output associated with a first deep network serving as a user-independent model associated with learning of one or more neural networks at the apparatus having the processor; and
learning on-the-fly logic to automatically generate training data for a second deep network serving as a user-dependent model, wherein the training data is generated based on the output,
wherein the learning on-the-fly logic is further to merge the user-independent model with the user-dependent model into a single joint model.

### Claim 2 (depends on 1)

The apparatus of claim 1, further comprising offline learning logic to perform offline training of the user-independent model associated with one or more of the neural networks, wherein the user-independent model represents a context-independent model.

### Claim 3 (depends on 1)

The apparatus of claim 1, wherein the learning on-the-fly logic is further to on-the-fly train the joint model to serve as a user-dependent classifier, wherein the user-dependent classifier represents a context-dependent classifier.

### Claim 4 (depends on 3)

The apparatus of claim 3, wherein training is performed in runtime and during normal operations of the apparatus, wherein the normal operations include one or more of launching or running of an application.

### Claim 5 (depends on 3)

The apparatus of claim 3, wherein training the joint model comprises mediating between one or more layers of the user-independent model and one or more layers of the user-dependent model that are converged within the joint model.

### Claim 6 (depends on 1)

The apparatus of claim 1, further comprising:
tuning logic to perform joint model tuning of one or more parameters of the joint model, wherein the joint model tuning is performed using small labeled data; and
generation and mediation logic to generate one or more mediation parameters to define mediation for learning or training of the joint model.

### Claim 7 (depends on 1)

The apparatus of claim 1, wherein the processor comprises a graphics processor co-located with an application processor on a common semiconductor package.

### Claim 8

A method comprising:
detecting an output associated with a first deep network serving as a user-independent model associated with learning of one or more neural networks at a computing device having a processor coupled to memory;
automatically generating training data for a second deep network serving as a user-dependent model, wherein the training data is generated based on the output; and
merging the user-independent model with the user-dependent model into a single joint model.

### Claim 9 (depends on 8)

The method of claim 8, further comprising performing offline training of the user-independent model associated with one or more of the neural networks, wherein the user-independent model represents a context-independent model.

### Claim 10 (depends on 8)

The method of claim 8, further comprising on-the-fly training the joint model to serve as a user-dependent classifier, wherein the user-dependent classifier represents a context-dependent classifier.

### Claim 11 (depends on 10)

The method of claim 10, wherein training is performed in runtime and during normal operations of the apparatus, wherein the normal operations include one or more of launching or running of an application.

### Claim 12 (depends on 10)

The method of claim 10, wherein training the joint model comprises mediating between one or more layers of the user-independent model and one or more layers of the user-dependent model that are converged within the joint model.

### Claim 13 (depends on 8)

The method of claim 8, further comprising:
performing joint model tuning of one or more parameters of the joint model, wherein the joint model tuning is performed using small labeled data; and
generating one or more mediation parameters to define mediation for learning or training of the joint model.

### Claim 14 (depends on 8)

The method of claim 8, wherein the processor comprises a graphics processor co-located with an application processor on a common semiconductor package.

### Claim 15

At least one machine-readable medium comprising instructions that when executed by a computing device, cause the computing device to perform operations comprising:
detecting an output associated with a first deep network serving as a user-independent model associated with learning of one or more neural networks, wherein the computing device includes a processor coupled to memory;
automatically generating training data for a second deep network serving as a user-dependent model, wherein the training data is generated based on the output; and
merging the user-independent model with the user-dependent model into a single joint model.

### Claim 16 (depends on 15)

The machine-readable medium of claim 15, wherein the operations further comprise performing offline training of the user-independent model associated with one or more of the neural networks, wherein the user-independent model represents a context-independent model.

### Claim 17 (depends on 15)

The machine-readable medium of claim 15, wherein the operations further comprise on-the-fly training the joint model to serve as a user-dependent classifier, wherein the user-dependent classifier represents a context-dependent classifier.

### Claim 18 (depends on 17)

The machine-readable medium of claim 17, wherein training is performed in runtime and during normal operations of the apparatus, wherein the normal operations include one or more of launching or running of an application.

### Claim 19 (depends on 17)

The machine-readable medium of claim 17, wherein training the joint model comprises mediating between one or more layers of the user-independent model and one or more layers of the user-dependent model that are converged within the joint model.

### Claim 20 (depends on 15)

The machine-readable medium of claim 15, wherein the operations further comprise:
performing joint model tuning of one or more parameters of the joint model, wherein the joint model tuning is performed using small labeled data; and
generating one or more mediation parameters to define mediation for learning or training of the joint model, wherein the processor comprises a graphics processor co-located with an application processor on a common semiconductor package.