Synthetic data generation in computer-based reasoning systems

1. A method comprising:

receiving a request for generation of synthetic training data based on a set of training data cases;

determining one or more focal training data cases from among the set of training data cases;

for each undetermined feature in the one or more focal training data cases,

determining a value for the undetermined feature in a synthetic data case based at least in part on the focal training data cases;

determining whether the value for the undetermined feature is valid based on feature information;

when it is determined that the value for the undetermined feature is invalid, performing a corrective action for the value of the feature;

when it is determined that the value for the undetermined feature is valid, using the value for the undetermined feature; and

causing control of a controllable system using a computer-based reasoning model that was determined at least in part based on the synthetic data case;

wherein the method is performed by one or more computing devices.

2. The method of claim 1 , wherein determining whether the value for the undetermined feature is valid based on feature information comprises:

determining whether the value for the undetermined feature is valid based on a comparison of the value with feature bounds.

3. The method of claim 2 , wherein determining whether the value for the undetermined feature is valid based on feature information comprises:

determining the feature bounds based on values of one or more other features in the synthetic data case.

4. The method of claim 2 , wherein performing the corrective action comprises:

determining a new value for the undetermined feature based on a distribution between the determined feature bounds.

5. The method of claim 1 , wherein performing the corrective action comprises:

determining a new value for the undetermined feature in the synthetic data case based at least in part on a distribution associated with the undetermined feature in the focal training data cases.

6. The method of claim 1 , wherein determining the value for the undetermined feature in the synthetic data case comprises:

determining the value for the undetermined feature based at least in part on a distribution associated with the undetermined feature.

7. The method of claim 1 , wherein the controllable system is a self-driving vehicle, the method further comprising:

receiving a current context for the self-driving vehicle during operation of the self-driving vehicle;

determining a suggested action for the self-driving vehicle based at least in part on the synthetic data case and the current context for the self-driving vehicle;

wherein causing control of the controllable system comprises causing performance of the suggested action by the self-driving vehicle.

8. The method of claim 1 ,

wherein the request includes a target amount of surprisal for the synthetic training data; and

wherein determining the value for the undetermined feature in the synthetic data case comprises determining the value for the undetermined feature in the synthetic data case based at least in part on a distribution associated with the undetermined feature in the focal training data cases and the target amount of surprisal.

9. The method of claim 8 , wherein the request includes the target amount of surprisal for the synthetic training data; and wherein determining, for each undetermined feature in the one or more focal training data cases, the value for the undetermined feature in the synthetic data case comprises:

determining a portion of the target amount of surprisal attributable to the undetermined feature by splitting surprisal evenly among the undetermined features, where each of N undetermined features uses (target surprisal)/N amount of the surprisal; and

determining the distribution for the undetermined feature based on the portion of the target amount of surprisal attributable to the undetermined feature.

10. The method of claim 1 , further comprising:

determining a fitness score for the synthetic data case; and

when the fitness score for the synthetic data case is beyond a particular threshold, using the synthetic data case as synthetic training data.

11. The method of claim 1 , further comprising:

determining a shortest distance between the synthetic data case and cases in the set of training data cases; and

when the shortest distance between the synthetic data case and the cases in the set of training data cases is beyond a particular threshold, using the synthetic data case as synthetic training data.

12. The method of claim 1 , further comprising:

determining distances between the synthetic data case and at least two cases in the set of training data cases;

determining whether there are at least a certain number (k) of training data cases that have a distance to the synthetic data case that is below a threshold; and

when there are at least k training data cases that have distances to the synthetic data case that are below the threshold, using the synthetic data case as synthetic training data.

13. The method of claim 1 , further comprising:

determining distances between the synthetic data case and at least two cases in the set of training data cases;

determining whether there are at least a certain number (k) of training data cases that have a distance to the synthetic data case that is below a first threshold;

determining whether any of the training data cases have a distance below a second threshold; and

when there are at least k training data cases have distances to the synthetic data case that are below the first threshold and no training data case has a distance to the synthetic data case that is below the second threshold, using the synthetic data case as synthetic training data.