AI Solution: DEEP:PHI
Core Technology: Pipeline technology, neural network modularization technology, artificial intelligence learning and operation optimization technology, etc.
The artificial intelligence platform is a concept that has emerged as interest in artificial intelligence analysis increases. Our DEEP:PHI is a platform that helps medical professionals with a simplified system so that they can directly analyze medical artificial intelligence if they only have medical images and data. In addition to providing computing resources in the form of cloud, it is also linked with DEEP:STORE, our artificial intelligence marketplace product, so that you can easily register your own solutions.
DEEP:PHI is a web-type platform that allows users to upload data and perform data preprocessing, artificial intelligence learning, and clinical verification through a modular pipeline. Each module can be used in the form of a GUI (Graphical User Interface) without the need for programming knowledge or coding, so it can be easily used even if you are not a programming expert. In addition, by providing various sample projects in addition to the provided manuals and video tutorials, even users without basic knowledge of artificial intelligence can easily understand and use them with only a short time training. It is suitable for explaining the step-by-step schematic diagram and the big picture of artificial intelligence, so it can be used for artificial intelligence education purposes.
DEEP:PHI is a platform that enables intuitive AI model design without coding through GUI-based pipeline technology. Since various image processing modules and neural network modules are provided as standard, medical AI models can be completed through a simple connection process. It reports AI learning results in various ways and provides the function to collaborate with other researchers if necessary. All projects and modules of DEEP:PHI can be shared and used as a platform to increase synergy of research.
A. Application of GUI based Pipeline concept without coding
DEEP:PHI is a platform that enables intuitive AI model design without coding through GUI-based pipeline technology. It utilizes pipeline technology that integrates and manages all configuration information, data sets, model storage, result information, and prediction information within a project within a graphical user interface. It visualizes the research results graphically to the user, and enables model configuration by dragging and dropping icons.
B. Free high-performance Neural Network module
Configuring a neural network is a very complex and time consuming task because each user has a different programming language, framework, etc. We basically provide high-performance neural network modules (ResNet, Inception, U-Net, DeepLab, etc.) that have been developed, and you can modify the network with the neural network model builder. In addition, it also supports the ability for users to create networks.
Image processing module is a modularized technology of advanced image processing and data processing used before artificial intelligence learning. The image processing module can also be used by calling the module and connecting it directly, and it can be used without coding as the parameter values of the module can be modified. If detailed modification is required, you can open an additional command window and modify the code.
C. Visualization of artificial intelligence research performance indicators
It is a technology that automatically visualizes the results of artificial intelligence learning and provides them to users. It shows the overview of the whole learning, the learning result, and the flow structure that was configured for learning, and the user can selectively examine each performance index for each learning session and target in detail.
D. Project collaboration support between users
Project sharing and collaboration functions are supported within the platform when you want to collaborate with someone who is conducting research similar to yourself. You can grant privileges and roles by providing a management function to add/delete members in the project. It is designed so that you can freely and conveniently conduct research projects within the platform without using separate collaboration tools.
E. Providing optimal AI research environment through cloud computing
When building a server system directly for artificial intelligence research and services, in order to increase service safety and provide convenience to users, a high-end server system with high-performance GPU must be built. In addition to expensive expenses, it is true that there is a need for professionals who can operate special high-performance servers, which is a burden.
DEEP:PHI utilizes cloud computing technology to provide an optimal research environment. In particular, for image-based artificial intelligence research, resource allocation management technology is essential because each user uses a lot of resources such as CPU memory, GPU memory, and storage. We provide an automated management system to efficiently operate such large-scale systems and quickly respond to failures.