The practices of data-driven marketing and risk control are employed by virtually all financial institutions today. Amid the diversity of financial products, scenarios, and data, the primary challenge is to enhance the timeliness of modeling for AI-enabled banking. Therefore, to facilitate enterprise-grade marketing and risk-control modeling, it is essential to build a machine learning platform that offers comprehensive functionality, such as advanced mathematics, AI algorithms, and high-performance computing. A professional AI platform of that kind requires not only secure, powerful, sharable, and easy-to-use AI tools, but also a model life-cycle management system and a wealth of available talent. As a consequence, the long-term investment in personnel training and the risk of turnover are inevitable.
The AI development platform, Model Magic, is a one-stop AI solution for data, feature engineering, modeling, and deployment reasoning. Its production-support capabilities for enterprises encompass model version management, iteration training, and life cycle management. Moreover, the complete algorithm operators ensure a pleasant user experience. Model Magic, thanks to years of careful development, now provides support for AI business scenarios at more than 20 banks.
Financial industry solution: Provide vertical application programs for the financial industry, such as credit-card installment, financial product recommendation, precision marketing, and customer churn alert so as to help users pursue their business goals without the need for long-term training and learning.
Open and secure: Fully independent intellectual property rights, open-source and open technological framework are compatible with China's proprietary platforms or X86 open hardware architecture.
Full-stack machine learning capabilities: Comprehensive and up-to-date statistical methods that are available to support diverse data structures. One-stop data analysis covers all phases of data handling.
Productivity improvement: The AI AutoML core technology is used to automate data exploration, preprocessing, feature engineering, and algorithm tuning.
Intensive sharing of computing power: A platform carries multiple businesses. Build an end-to-end machine learning application platform to support the applications of each branch in the financial industry.
Designed for business scenarios in financial risk control, this function combines risk-control modeling, scorecards, and many more features. It also facilitates the quantitative assessment of pre-lending, lending, and post-lending risk control and fraud detection.
AI recommendation and precision marketing are based on the thorough understanding of customers and products as well as in-depth analysis of behavioral characteristics, thereby establishing an AI model that supports the ecology of digital marketing.
Support NLP and CV multi-modal AI human-computer interaction as the basic platform while also facilitating smart customer service, smart conversation, and self-service banking.
Deep learning capabilities of the platform are integrated with RPA and widely used in large-scale smart operation scenarios such as smart billing and automated financial reimbursement.
Create platform-based visual applications that encompass paired digital scenarios such as Internet content auditing, video recognition and analysis, together with the digital virtual human.
Six technical features: Focus on the sustainability of AI production and application:
A platform for one-stop data analysis, mining, and deployment.
ML/DL automated parameter tuning and optimization as well as quick modeling.
03Visualization and coding
The entire modeling process is accomplished with drag-and-drop simplicity. Bottom-level programming is also supported.
04Full-process model management
Full-process model management, data processing and training, deployment integration, and provision of public datasets and case libraries.
05Open AI framework
Support for custom algorithms is provided along with hundreds of classic machine learning and deep learning algorithms.
Distributed computing, data parallelism, and algorithmic parallelism.