FIELD: computer engineering.
SUBSTANCE: result is achieved due to steps of: obtaining a text description of a test object to form its test image, wherein the textual description in the natural language indicates what should be displayed on the test image; obtaining keywords related to text description, each of which indicates a rendering command for displaying a test object on a test image; formation of augmented text descriptions of the image based on the key words; inputting into the GMLM model each augmented text description to form candidate images of the object; transmitting candidate images to a plurality of human evaluators for pairwise comparison thereof; determining the degree of visual attractiveness for the candidate image based on the pairwise comparison results; and using the degree of visual appeal to refine the GMLM.
EFFECT: high visual attractiveness of images formed by means of a pre-trained generative machine learning model (GMLM).
20 cl, 8 dwg
| Title | Year | Author | Number |
|---|---|---|---|
| SYSTEM AND METHOD FOR TRAINING MACHINE LEARNING MODELS FOR RANKING SEARCH RESULTS | 2023 |
|
RU2829065C1 |
| METHOD AND SYSTEM FOR CHECKING MEDIA CONTENT | 2022 |
|
RU2815896C2 |
| METHOD AND SYSTEM FOR RANKING DIGITAL OBJECTS BASED ON TARGET CHARACTERISTIC RELATED TO THEM | 2019 |
|
RU2757174C2 |
| METHOD AND SYSTEM FOR GENERATING DIGITAL TASK LABEL BY MACHINE LEARNING ALGORITHM | 2022 |
|
RU2829151C2 |
| METHOD AND SERVER FOR TEACHING A NEURAL NETWORK TO FORM A TEXT OUTPUT SEQUENCE | 2020 |
|
RU2798362C2 |
| METHOD AND SYSTEM FOR DETERMINING DEGREE OF DETERIORATION OF LIDAR DATA | 2021 |
|
RU2826476C1 |
| METHOD AND SYSTEM FOR PROCESSING USER SPOKEN SPEECH FRAGMENT | 2019 |
|
RU2757264C2 |
| MULTISTAGE TRAINING OF MACHINE LEARNING MODELS FOR RANKING SEARCH RESULTS | 2021 |
|
RU2824338C2 |
| METHODS AND SYSTEMS FOR FORMING TRAINING DATA FOR A NEURAL NETWORK | 2020 |
|
RU2763215C2 |
| METHOD AND SYSTEM FOR CLASSIFYING AN ELECTRONIC DEVICE USER | 2021 |
|
RU2795152C2 |
