Apparatus and method for database management of machine learning models
AssigneeSubstrate Artificial Intelligence SA
InventorJames Brennan Worth
Life of this patent
Estimating the term…
Abstract
In an embodiment, a method includes updating a first model storage that includes a first plurality of machine learning (ML) models to generate a second model storage that includes a second plurality of ML models. The second plurality of ML models is different than the first plurality of ML models. A representation of a request is received. A set of attributes associated with the request is identified. The second model storage is filtered, based on the set of state features, to identify a subset of ML models included in the second model storage. Each model from the subset of ML models is associated with at least one attribute from the set attributes. The request is caused to be processed using the subset of ML models to generate an output.
Claims22
A method, comprising:
- receiving a representation of a request;
- identifying a set of attributes associated with the request;
- updating a first model storage that includes a first plurality of machine learning (ML) models to generate a second model storage that includes a second plurality of ML models, the second plurality of ML models including an ML model that is not included in the first plurality of ML models, the set of attributes represented using a set of state features and the ML model trained using a process that includes:
- receiving training data;
- validating the training data;
- randomizing, after validating the training data, the training data to generate randomized training data;
- and
- training the ML model using the randomized training data;
- filtering the second model storage, based on the set of attributes, to identify a subset of ML models included in the second model storage, the subset of ML models including multiple ML models from the second model storage, each ML model from the subset of ML models associated with at least one attribute from the set of attributes;
- and
- causing, to generate an output, the request to be processed using the subset of ML models.
A non-transitory processor-readable medium storing code representing instructions to be executed by a processor, the instructions comprising code to cause the processor to:
- receive a representation of a request;
- determine that a size associated with the request is larger than a predetermined threshold;
- divide the request into a plurality of parts, each part from the plurality of parts not larger than the predetermined threshold,
- generate a set of vectors based on the request and using the plurality of parts, a capacity of each vector from the set of vectors being less than the predetermined threshold, each part from the plurality of parts associated with at least one vector from the set of vectors;
- identify, based on the set of vectors and using a vector database that includes a plurality of vectors associated with a plurality of machine learning (ML) models, a subset of vectors from the plurality of vectors, the subset of vectors associated with a subset of ML models from the plurality of ML models, the subset of ML models including multiple ML models;
- and
- cause, to generate an output, the request to be processed using at least two ML models from the subset of ML models.
An apparatus, comprising:
- a memory;
- and
- a processor operatively coupled to the memory, the processor configured to:
- receive a representation of a request;
- identify at least one of a set of state features or a set of vectors associated with the request;
- filter a model storage that includes a plurality of machine learning (ML) models to identify a subset of ML models included in the model storage, each ML model from the subset of ML models associated with at least one of (1) at least one state feature from the set of state features or (2) at least one vector from the set of vectors, an ML model from the plurality of ML models trained using a process that includes:
- receiving training data;
- validating the training data;
- randomizing, after validating the training data, the training data to generate randomized training data;
- and
- training the ML model based on the randomized training data;
- and
- cause, to generate an output, the request to be processed using the subset of ML models and not remaining ML models from the model storage.
Dependencies are inferred from the claim text.