Systems and methods to improve data clustering using a meta-clustering model

1. A user device, comprising:

one or more memory units storing instructions;

a user interface configured to receive user input; and

one or more processors configured to execute the instructions to perform operations including:

transmitting, to a server, a clustering request, the request configured to cause the server to:

generate, using a plurality of embedding network layers, preliminary data clusters; and

transmit, to the user device for display, a request for user input;

displaying, at the user interface, the request for user input; and

receiving, at the user interface, a user input configured to cause the server to:

generate, using a meta-clustering model, final data clusters based on the preliminary data clusters; and

store the final data clusters.

2. The user device of claim 1 , wherein the request is further configured to cause the server to:

generate an updated embedding network by training a first embedding network layer of the plurality of embedding network layers; and

generate, using the updated embedding network, updated data clusters.

3. The user device of claim 1 , wherein the operations further comprise:

determining to request user input based on a predetermined command to perform supervised model training.

4. The user device of claim 3 , wherein the received user input identifies a data classification category.

5. The user device of claim 1 , wherein the received user input associates tags with data cluster samples.

6. The user device of claim 1 , wherein the operations further comprise receiving, at the user interface, an additional user input configured to cause the server to:

send, to the user device for display on the user interface, at least one of:

the final data clusters;

a visual representation of the final data clusters;

a number of clusters in the final data clusters; or

the meta-clustering model.

7. The user device of claim 6 , wherein the number of clusters is larger than a maximum count of clusters in the preliminary data clusters.

8. The user device of claim 1 , wherein the operations further comprise transmitting, to a server, a clustering request, the clustering request configured to cause the server to:

generate data cluster samples based on the preliminary data clusters; and

generate tags associated with the data cluster samples using the meta-clustering model.

9. The user device of claim 1 , wherein generating preliminary data clusters comprises:

receiving training data;

generating a first embedding network layer of the plurality of embedding network layers;

training the generated first embedding network layer to classify training data; and

training the generated first embedding network layer to cluster training data.

10. The user device of claim 9 , wherein generating preliminary data clusters comprises repeating generating steps, the generating steps comprising:

adding a trained embedding network layer to the plurality of embedding network layers;

generating data clusters using the added trained embedding network layer; and

tagging the data clusters.

11. The user device of claim 10 , wherein the generation of preliminary data clusters terminates when a performance criterion is satisfied.

12. The user device of claim 1 , wherein the preliminary data clusters comprises image data and the embedding network layer comprises a convolutional neural network.

13. The user device of claim 1 , wherein the preliminary data clusters comprises text data and the embedding network layers are part of a language representation model.

14. The user device of claim 1 , wherein the plurality of embedding network layers comprises at least one of a Bidirectional Encoder Representations from Transformers (BERT) model or an Embeddings from Language Models (ELMo) representation model.

15. The user device of claim 1 , wherein the received user input comprises a selection of a multi-clustering model.

16. The user device of claim 1 , wherein generating preliminary data clusters comprises performing at least one of a binary classification or one-hot encoding.

17. A method for clustering data, comprising:

transmitting, to a server, a clustering request configured to cause the server to:

generate, at the server using a plurality of embedding network layers, preliminary data clusters; and

transmit, to a user device for display, a request for user input;

displaying, at the user device, the request for user input; and

receiving, at the user device, a user input configured to cause the server to:

generate, using a meta-clustering model, final data clusters based on the preliminary data clusters; and

store the final data clusters.

18. The method of claim 17 , wherein the received user input associates tags with data cluster samples.

19. A user device for clustering data comprising:

one or more memory units storing instructions;

a user interface configured to receive user input; and

one or more processors configured to execute the instructions to perform operations including:

transmitting, to a server, a clustering request, the clustering request configured to cause the server to generate, using a plurality of embedding network layers, preliminary data clusters;

displaying, at the user interface, a request for user input received from the server; and

receiving, at the user interface, a user input configured to cause the server to:

generate, using a meta-clustering model trained using reduced-dimensionality data clusters, final data clusters based on the preliminary data clusters; and

store the final data clusters.

20. The user device of claim 19 , wherein training the meta-clustering model comprises iteratively repeating training steps until the performance criterion is satisfied, the training steps comprising:

determining, using the meta-clustering model, a number of clusters based on the data map;

generating an updated embedding network based on the number of clusters;

generating updated data clusters using the updated embedding network;

updating the meta-clustering model;

reducing a dimensionality of the updated data clusters;

generating, using the updated meta-clustering model, updated encoded data based on the reduced-dimensionality updated data clusters;

generating an updated data map based on the updated encoded data; and

determining whether the performance criterion is satisfied based on the updated data map.