SOLUTIONS

Beyond the Limits of the Actual Operating Environment with Deep Learning Solution Beyond Learning

1. Integrated Monitoring and Log Analysis System
  • Improve operational efficiency by automating service monitoring, detection alerts, and analysis reporting through new cluster deployments.
monitoring_01

<Development of Network Inspection Checklist Tools (2019)>

02_00

<Developing Verification Management/Comparison Analysis Tool for the whole vehicle (2018)>

03_00

<Develop Network Visualization Tools (2016)>

2. Deep Learning Based Braking Optimization System
  • Overcomes the limitations of the real operating environment through a deep learning based parameter optimization system.
(1) Generate Learning Data (2) Performing Deep Learning (3) Predicting and Optimizing Results
  • Automate learning data generation and storage processes using commercial CARSIM simulators
  • Minimize learning data generation period by developing scheduleers and interlocking API
  • Model as a deep enough network using the latest deep learning learning methods (ReLU, DropOut, Batch Normalization, etc.) and enable learning without overfitting (more than 90% Acuracy)
  • Quickly predict results with learned classification models
  • Optimize parameters in the same way as Genetic Algorithm if learned models are available
ml12
3. Development of Deep Learning Based Vehicle Recognition System
  • We have a variety of scalability features including vehicle recognition based on the Heat Map, Fake reading through the photo texture recognition, and damage identification using the deep learning algorithm.
__ml1234
Deep Learning Based Vehicle Recognition Demo