Data labeling for deep-learning models
A first and second scoring endpoint with payload logging are deployed. At the second scoring endpoint, native data and a user-generated score for the native data are received, the native data is pre-processed into readable data for the deep-learning model, and the user-generated score and the readable data are output to the first scoring endpoint, which is associated directly with the deep-learning model. A raw payload that includes the native data is output to a payload store. At the first scoring endpoint, the readable data and the user-generated score are processed by the deep-learning model, which outputs a transformed payload and a prediction, respectively, to the payload store. The raw payload is matched with the transformed payload and the prediction to produce a comprehensive data set, which is evaluated to describe a set of transformation parameters. The deep-learning model is retrained to account for the set of transformation parameters.
1. A system for managing deep-learning, comprising:
a memory with program instructions stored thereon; and
a processor in communication with the memory, wherein the program instructions cause the system to:
deploy a first and a second scoring endpoint with payload logging for a deep-learning model;
receive, at the second scoring endpoint, native data and a user-generated score for the native data;
pre-process, at the second scoring endpoint, the native data into readable data for the deep-learning model;
output, from the second scoring endpoint to the first scoring endpoint, the user-generated score for the native data and the readable data, wherein the first scoring endpoint is associated directly with the deep-learning model;
output, from the second scoring endpoint to a payload store, a raw payload, wherein the raw payload includes the native data;
process, at the first scoring endpoint and using the deep-learning model, the readable data and the user-generated score to output a transformed payload and a prediction, respectively, to the payload store;
match, at the payload store, the raw payload with the transformed payload and the prediction to produce a comprehensive data set;
evaluate the comprehensive data set to describe a set of transformation parameters; and
retrain the deep-learning model to account for the set of transformation parameters.
2. The system of claim 1 , wherein the program instructions further cause the system to:
evaluate, based on comparing the prediction and the transformed payload, the performance of the deep-learning model; and
use the performance evaluation of the deep-learning model to retrain the deep-learning model.
3. The system of claim 1 , wherein evaluating the comprehensive data set includes data labeling, according to the functions of the neurons of the deep-learning model.
4. The system of claim 3 , wherein the retraining includes multiple iterations of the deep-learning model.
5. The system of claim 4 , wherein the matching is performed using a unique scoring ID, wherein the unique scoring ID is assigned to the user-generated score and the raw data, and wherein the unique scoring ID is subsequently associated with the raw payload, transformed payload, and the user-generated score.
6. The system of claim 5 , wherein the native data includes a video file and the transformed payload includes an identification of an object depicted within the video.
7. The system of claim 6 , wherein software is provided as a service in a cloud environment to execute the program instructions.
8. A computer program product for managing deep-learning, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a device to cause the device to:
deploy a first and a second scoring endpoint with payload logging for a deep-learning model;
receive, at the second scoring endpoint, native data and a user-generated score for the native data;
pre-process, at the second scoring endpoint, the native data into readable data for the deep-learning model;
output, from the second scoring endpoint to the first scoring endpoint, the user-generated score for the native data and the readable data, wherein the first scoring endpoint is associated directly with the deep-learning model;
output, from the second scoring endpoint to a payload store, a raw payload, wherein the raw payload includes the native data;
process, at the first scoring endpoint and using the deep-learning model, the readable data and the user-generated score to output a transformed payload and a prediction, respectively, to the payload store;
match, at the payload store, the raw payload with the transformed payload and the prediction to produce a comprehensive data set;
evaluate the comprehensive data set to describe a set of transformation parameters; and
retrain the deep-learning model to account for the set of transformation parameters.
9. The computer program product of claim 8 , wherein the program instructions further cause the device to:
evaluate, based on comparing the prediction and the transformed payload, the performance of the deep-learning model; and
use the performance evaluation of the deep-learning model to retrain the deep-learning model.
10. The computer program product of claim 8 , wherein evaluating the comprehensive data set includes data labeling, according to the functions of the neurons of the deep-learning model.
11. The computer program product of claim 10 , wherein the retraining includes multiple iterations of the deep-learning model.
12. The computer program product of claim 11 , wherein the matching is performed using a unique scoring ID, wherein the unique scoring ID is assigned to the user-generated score and the raw data, and wherein the unique scoring ID is subsequently associated with the raw payload, transformed payload, and the user-generated score.
13. The computer program product of claim 12 , wherein the native data includes a video file and the transformed payload includes an identification of an object depicted within the video.