Brightics Deep Learning
An unstructured data analytics platform - Deep learning service made easy and fast for all
Recommend optimum models and label data automatically, for maximum productivity in deep learning development process
Major Services
Find the fastest training speed in the field of deep learning
- Save image labelling time 70% World's highest auto-labelling rate
- Improve model selection speed 2~8 times Automatic DL model recommendation
-
Automation of the labeling process for learning data
The labeling time will be reduced 70-80% thanks to the auto-labeling technology by Brightics Deep Learning. In addition, costs and trial & error will be reduced as well.
-
Recommend learning models in earlier stages
The learning pace will be improved by providing recommended functions of learning models in eariler stages, which is optimized for learning data.
-
Distributed process for data in bulk
The pace of data process in bulk will be 8 times more improved by securing high-performance distributed processing, which proceeds with collection of data in bulk/pre-process/learning processes in distributed manners.
Use Cases
-
ECG analytics with higher accuracy by utilizing deep learningBy doing deep learning analytics for ECG data collected through wearable devices, the data of sensitive and various types of arrhythmia can be accurately diagnosed. Hospitals and research institutes can utilize such analytics results for treatment and clinical research. -
Predict risk levels of diseases and provide early disease prevention by analyzing genome dataIt is possible to check whether those with risks of hereditary cancers (e.g. family medical history) were actually born to have genotypic variation of cancers by thoroughly reviewing the list of targeted genotypic variation provided by Genomics in a clinical manner. -
Auto-quotation for repairing by anlyzing images of car accidentsThe duration for insurance payment process can be reduced and proper insurance premium can be calculated by providing proper repairing and insurance premium through pattern analysis of accident images in the case of processing car accidents.
Features
-
High-performance distributed process engine
The pace of data process in bulk will be 8 times more improved by securing the world's best high-performance distributed processing, which applies the most optimized load distributed methods to the company's own patented DB and the analytics server.
-
Best Model Proposal
The time that requires on-site experts or less skilled AI developers to select deep learning models will be at least 2-8 times more improved by utilizing the pre-trained model that automatically recommends the most optimized deep learning models for the given learning data.
-
Auto-Labeling
The labeling time will be reduced up to 80% by automatically carrying out the 80% of data labeling while human can implement the rest of 20% by maximizing the labeling automation rate with the most advanced active learning technology.
Specifications
-
Deep learning analytics platform
x86 server with 2 or more nodes
-GPU: NVIDIA V100 1EA/Node
-CPU: 16 cores/Node
-Memory: 32GB/Node
-Operating system: Linux 64bits -
PC for platform users
- Browser : Chrome (Version 50.0 or higher)
- Screen resolution: 1280 x 900 (recommended) -
Appropriate cloud environments
SDS Cloud, Amazon Web Service, MS Azure, Google Cloud, etc.
Documentations
Related Offerings
Find out more about our offerings
Brightics ML is an AI-based data analytics platform for implementing quick and efficient analytics services within an enterprise. It provides an environment for visualized analytics modeling, as well as support for simple script development.
Brightics DP is a data management platform that supports efficient preparation of analysis data. It allows enterprise users to quickly collect and utilize data generated internally and externally for their analytics services.
Distributed Data Platform - Optimized Hadoop Ecosystem for Enterprise Business
Whether you’re looking for a specific business solution or just need some questions answered, we’re here to help
