ESD KG4208-G4 empowers an education-focused AI company to build an intelligent teaching system

 

Background

A Beijing-based AI company specializing in education released its first public online project platform in 2014, connecting global talents, enterprises, and universities through internet technology and exploring new models for future education. In three years, hundreds of thousands of young people have been involved in hundreds of high-quality projects, with an impact spreading to 321 cities in 17 countries worldwide.

 

In 2017, the company received a multi-million-dollar strategic investment from a listed A-share company, LSC Group, and shifted its development direction to "AI+education". The company aims to provide comprehensive, fast, and fully automated classroom teaching analysis through AI technologies such as emotion computing, identity recognition, and natural language understanding, and provide solutions and technical services for offline classroom teaching, online one-on-one teaching, and professional training scenarios.

 

Core Technologies

Facial expression recognition

 

The movement of facial muscles has a certain regularity when expressing the same emotions, regardless of gender, age, race, and education level. In classroom teaching, changes in students' facial expressions can directly reflect the degree of acceptance of teaching content and their preference for the teacher's teaching style.

 

Speech emotion recognition

Changes in the autonomic nervous system can indirectly affect a person's speech. Emotional speech processing technology can use speech feature analysis to recognize human emotional states. In the teaching scenario, teachers and students tend to have wider and higher vocal ranges when they are in a pleasant state. In contrast, in emotions such as fatigue and confusion, speech tends to become slow, low, and unclear. Speech analysis can efficiently analyze the emotional state of the classroom.

 

Classroom emotion report

The teaching analysis system is based on the analysis of teachers' and students' subtle emotional changes in classroom teaching videos, combined with teaching methodology for in-depth research. The results are presented visually to teaching managers, teachers, and parents in the form of learning reports, which help teachers optimize their teaching methods and enhance students' learning experiences.

 

Core Computing Power

"Facial expression recognition" and "speech emotion recognition" require the training of intelligent neural networks from massive facial expression and speech databases, which cannot be achieved without supercomputing. The ESD KG4208-G4 is a high-performance server designed by ESD Technology specifically for AI and deep learning, using GPU parallel computing technology to accelerate the training of neural networks.

 

ESD KG4208-G4 Features

Supports 4 NVIDIA GPU accelerators to enhance neural network training

Supports 2 Intel Xeon E5-2600 processors and 512GB of memory

Supports 8 SATA/SSD hard drives, Intel dual Gigabit Ethernet

4U rack/tower specifications, supports 1+1 redundant power supply

Unique GPU cooling design to improve system reliability

 

Result

AI technologies such as facial expression recognition, speech emotion analysis, and natural language understanding help teachers grasp more than 20 dimensions of information to gain a more comprehensive understanding of students' characteristics and develop personalized learning plans down to the knowledge point. Providing parents with timely and comprehensive learning reports helps them understand their children's learning situation, adjust their education methods, and detect and solve problems in a timely manner. This promotes the collaboration of schools, teachers, and parents to provide students with better teaching environments, more personalized learning guidance, and more considerate learning care.