Systems and methods for distributed data analytics
The invention provides systems and method for generating device-specific artificial neural network (ANN) models for distribution across user devices. Sample datasets are collected from devices in a particular environment or use case and include predictions by device-specific ANN models executing the user devices. The received datasets are used with existing datasets and stored ANN models to generate updated device-specific ANN models from each of the stored instances of the device ANN models based on the training data.
1 . A method for generating device-specific artificial neural network (ANN) models for distribution across a plurality of user devices, the method comprising:
receiving, by a processor, sample datasets for object recognition from at least two of the plurality of user devices,
writing, by the processer, the sample datasets to a first data store;
identifying, by the processor and in a second data store, a use case dataset for object recognition within an environmental setting, the use case dataset comprising at least one or more training data parameters and an aspect of the environmental setting in which a user device of the plurality of user devices is placed;
identifying, by the processor and in the first data store, training data from the sample datasets that meet the one or more training data parameters;
identifying, by the processor and in the second data store, stored instances of the device-specific ANN models; and
generating, by the processor, updated device-specific ANN models from each of the stored instances of the device ANN models based on the training data, wherein each of the updated device-specific ANN models are specific to a hardware component and environment of a corresponding user device of the at least two of the plurality of user devices.
2 . The method of claim 1 , wherein the sample datasets comprise media data and predictions by device-specific ANN models executing on the user devices.
3 . The method of claim 2 , wherein the media data comprises image data and application of the ANN models to the image data facilitates the identification of objects of interest within the image data.
4 . The method of claim 2 , wherein the one or more training data parameters include media data parameters and device parameters.
5 . The method of claim 4 , wherein the media data parameters include one or more of a color index, a brightness index, a contrast index, an image temperature, a tint, one or more hue values, and a gamma value.
6 . The method of claim 4 , wherein the device parameters include one or more of available memory, processing speed, image resolution, and frame-rate capture.
7 . The method of claim 1 , wherein the user devices comprise a plurality of image capture devices.
8 . The method of claim 1 , wherein the use case dataset is specific to a particular use case.
9 . The method of claim 8 , wherein a use case comprises a functional aspect.
10 . The method of claim 9 , wherein the functional aspect of the use case comprises facial recognition.
11 . The method of claim 1 , wherein the environmental aspect of the use case comprises one of placement of the device in an outdoor environment, placement of the device in an indoor environment, placement of the device in a well-lit environment, or placement of the device in a poorly-lit environment.
12 . The method of claim 1 , wherein the predictions generated by the device-specific ANN models comprise a quantitative image salience metric for each image that indicates the likelihood that the media file contains an object of interest.
13 . The method of claim 12 , further comprising determining, based at least in part on the quantitative image salience metrics for each image, a minimum number of images necessary to achieve a threshold model accuracy.
14 . The method of claim 1 , further comprising maintaining a library of device-specific parameters and training data, and wherein generating the updated device-specific ANN models is further based on the device-specific parameters and training data.
15 . The method of claim 1 , further comprising distributing, by the processor, the device-specific updated ANN models to at least a subset of the user devices associated therewith.
16 . A system for generating device-specific artificial neural network (ANN) models for distribution across a plurality of user devices, the system comprising:
one or more processors; and
a memory coupled with the one or more processors wherein the one or more processors
executes a plurality of modules stored in the memory and wherein the plurality of modules
comprises:
a user interface that receives instructions from a user, the instructions identifying
one or more sample datasets from at least two of the plurality of user devices of a user environment;
a data store comprising the sample datasets;
a business logic module that, when executed, (i) identifies a use case dataset for object recognition within an environmental setting stored in the data store, the use case dataset comprising at least training data parameters and an aspect of the environmental setting in which a user device of the plurality of user devices is placed,
(ii) identifies training data from the sample datasets that meet the training data parameters provided in the use case dataset and (iii) identifies device-specific ANN models stored in the data store; and
an artificial intelligence machine learning module that when executed generates
updated device-specific ANN models from each of the stored instances of the device
ANN models based on the training data, wherein each of the updated device-specific ANN models are specific to a hardware component and environment of a corresponding user device of the at least two of the plurality of user devices.
17 . The system of claim 16 , wherein the sample datasets comprise media data and predictions by device specific ANN models executing on the user devices.
18 . The system of claim 17 , wherein the user devices comprise a plurality of image capture devices.
19 . The system of claim 17 , wherein the media data comprises image data and application of the ANN models to the image data facilitates the identification of objects of interest within the image data.
20 . The system of claim 17 , wherein the training data parameters include media data parameters and device parameters.
21 . The system of claim 20 , wherein the media data parameters include one or more of a color index, a brightness index, a contrast index, an image temperature, a tint, one or more hue values, and a gamma value.
22 . The system of claim 20 , wherein the device parameters include one or more of available memory, processing speed, image resolution, and frame-rate capture.
23 . The system of claim 16 , wherein the use case dataset is specific to a particular use case.
24 . The system of claim 23 , wherein a use case comprises a functional aspect.
25 . The system of claim 24 , wherein the functional aspect of the use case comprises facial recognition.
26 . The system of claim 16 , wherein the environmental aspect of the use case comprises one of placement of the device in an outdoor environment, placement of the device in an indoor environment, placement of the device in a well-lit environment, or placement of the device in a poorly-lit environment.
27 . The system of claim 17 , wherein the predictions generated by the device-specific ANN models comprise a quantitative image salience metric for each image that indicates the likelihood that the media file contains an object of interest.
28 . The system of claim 27 , wherein the artificial intelligence machine learning module further determines, based at least in part on the quantitative image salience metrics for each image, a minimum number of images necessary to achieve a threshold model accuracy.
29 . The system of claim 17 , further comprising a library of device-specific parameters and training data, and wherein the artificial intelligence machine learning module generates the updated device-specific ANN models based on the device-specific parameters and training data.
30 . The system of claim 17 , further comprising a deployment module for distributing the device-specific updated ANN models to at least a subset of the user devices associated therewith.
31 . A non-transitory computer-readable storage medium storing computer-readable instructions that, when executed by a computer, cause the computer to perform a method, the method comprising:
receiving, by a processor, sample datasets from at least two of a plurality of user devices,
writing, by the processer, the sample datasets to a first data store;
identifying, by the processor and in a second data store, a use case dataset for object recognition within an environmental setting, the use case dataset comprising at least one or more training data parameters and an aspect of the environmental setting in which a user device of the plurality of user devices is placed;
identifying, by the processor and in the first data store, training data from the sample datasets that meet the one or more training data parameters;
identifying, by the processor and in the second data store, stored instances of the device-specific ANN models; and
generating, by the processor, updated device-specific ANN models from each of the stored instances of the device ANN models based on the training data, wherein each of the updated device-specific ANN models are specific to a hardware component and environment of a corresponding user device of the at least two of the plurality of user devices.