Automated Machine Learning Platform
Conventionally, the classification and prediction tasks based on business big data heavily rely on engineers/researchers’ subjective experience in algorithm selection, feature engineering and hyperparameter optimization. Our automated machine learning platform can minimize such reliance. It integrates the whole or partial automatic data analysis, feature engineering, algorithm selection and hyperparameter optimization into the highly customizable interactive workflow, thus significantly decreases the cost of design, development, deployment and maintenance.
Content Recommendation Engine
No matter for online content distribution, entertainment community, e-commerce and advertising platform, or for offline scenario/user-oriented retail and marketing, big-data based intelligent recommendation engine can be the key driving force to increase the click-through rate and conversion rate, and to decrease the operation cost. Our intelligent recommendation engine integrates intelligent user persona, CTR prediction, A/B testing, overall ROI optimization and real-time model upgrading. It’s also compatible to the new generation of recommended algorithms, such as deep learning and reinforcement learning.
How to set up a computer vision database for hundreds of, thousands of or even more commodities in a short time? How to upgrade the system vision model quickly as a new product comes onto the shelf? How to precisely and efficiently recognize each commodity from images or videos in the environments of different light levels, color temperatures or noises. How to plan the deployment of the image acquisition unit, computing unit, communication unit and etc in different commercial environments, so to minimize the total cost of ownership? Our computer graphics and computer vision team doesn’t only have the frontier algorithms, but also have the rich experience of solving engineering problems.
Human Behavior Recognition
Fine-grained and high frame rate real-time recognition of human behavior is one of the most challenging tasks in computer vision. Our computer graphics and computer vision team provides algorithm modules and overall solutions developed with frontier technologies, such as facial expression recognition, gesture or posture recognition, human skeletal system keypoint recognition, behavior and action recognition, focus recognition, individual recognition as well as crowd behavior statistic and monitoring. These will help business acquire precise data of consumer behaviors and significantly improve the digital marketing and the customer relationship management capability.
Leveraging the latest deep neural network and model compression technology, our scene recognition algorithm can run efficiently in an embedded system or a mobile device, providing the content understanding capacity for real-time images and videos. Our SDK enables mobile APPs, AR applications and intelligent robots to quickly build the preliminary understanding capability of the environment.
Natural Language Understanding
Speech Intention and Emotion Recognition
Leveraging the new machine learning technology and the semantic retrieval ability of the cross-domain entity relationship database, we have developed a toolkit to analyze and classify speakers’ real intention and current emotion. The capability of more precise intention and emotion recognition makes the most important analysis and decision-making unit in the "brain" of the next generation of intelligent home appliances, smartphone assistant and intelligent vehicles, and the core of the future intelligent life.
We develop customized conversational system for your specific scenario and domain needs. This new generation of human-machine interactive interface will bring you the ease of use and scalability like never before.
Domain Knowledge Graph
We have the core experience of building the world's largest knowledge graph. We’re committed to apply the representation and reasoning technology in the knowledge graph field flexibly onto our clients' business, and build complete and efficient knowledge system for any specific business process or knowledge area.
Public Opinion Analysis
We acquire a client's business-related public opinion information from open internet or an authorized network, real-time and efficiently. The new generation of natural language understanding technology has enabled the information extraction and machine learning algorithm to gain the most vital information for business promotion from the massive data immediately.