We strive to advance AI technologies to improve life!
The human brain is a giant graph of 100 billion nodes and 700 trillion edges, performing four major functions: memory, observation, judgement & perception and abstract reasoning & strategy. Graphs are our natural way to remember, correlate and understand all that surrounds us. To mimic the human experience of intelligence, graphs play a key role.
For over two decades, our researchers and developers have been devoting themselves to advance artificial intelligence technologies based on the potential capabilities of the graph infrastructure. For instance, we can use them for:
Artificial Intelligence is making fundamental paradigm shift in industries. Graphen is currently providing services, solutions, and products in the Finance, Medical, Automotive, Energy and Security industries, based on a 'Full Brain' and its derivative brains of deep industry knowledge.
Named after Ardi, an early human-like female anthropoid 4.4 million years old - and currently considered the first human being, our AI foundation is trying to mimic all functions of the human brain. The differences between humans and other animals mainly lie in our capabilities of reasoning, understanding and strategizing.
Graphen Ardi foundation includes 11 toolkits:
Memory, Perception, Learning, Understanding, Reasoning, Strategy, Expression,
Personality, Emotion, Knowledge, and Action.
These modules can be coordinated by a Pipeline.
With the powerful graph-enabled Ardi AI Platform, Graphen Alpha collects large-scale of market and company information in the major markets. It estimates market treads through Reasoning of hundreds of indicators in corporate, macroeconomics, geospatial, and industry, with real-time market data. Through Alpha's large-scale Understanding, and Learning, its Knowledge is stored in the platform and powers financial advisor with Personality.
Graphen "AI Tools for Medicine (Atom)"" platform includes tools of four categories, powered by advanced AI Graph technology:
All of our tools' accuracy outperformed the best known tools in the literature, except the Protein Structure Prediction tool, in which our tool's accuracy is slightly worse but with less computational requirements.
Graphen Atom also incorporates Ardi Understanding to digest information from massive articles in the literature. A major outcome is to detect Ranked Health Risks based on static and dynamic Genome.
Graphen Aica is a platform full of automotive domain knowledge, including how the cars function and what to expect from the information gathered from car sensors. The platform also includes the makes, models, and parts information from various markets in the world. It can also decode the VINs in the world's major markets.
Aica successfully achieved close to 99% accuracy of car diagnosis and repair suggestions.
Buit upon Ardi Understanding, Learning, Reasoning, and Knowledge, Aica also listens from the drivers or technicians, examining sensors and diagnostic signals from the car computers through ELM327 compatible bluetooth device, and providing answers of general automotive or specific car's questions via human-like conversations.
Built with Ardi's advanced functionalities, Graphen Agni incorporates multimodality data to predict green energy production and consumption, such as using Ardi Perception and Ardi Reasoning to analyze the weather data and the grid workload.
Some of Graphen Agni's capabilities include:
Graphen Alice is an infrastructure that incorporates Ardi Reasoning to interpret machine or human behavioral anomalies. The key differentiator of Alice is its capability to detect anomalies from small signals, understand the steps the penetrators likely are, and thus predict what may be the next steps the intruders may act. Ardi Reasoning can precisely reveal the causality of abnormal behaviors. This makes "Pre-Crime" detection a possible reach.
Most entities are connected. A novel AI infrastructure can handle billions of connections like our brains.
Visualization is the key to understanding analysis. We need to combine data mining, big data analytics, design, rendering and interactions.
Continuous Imperfect Learning allows machines to self-learn its environments and keep improving itself, without explicit teachers.
Machines may evolve to have humanlike capabilities such as feeling images, reading body language, capturing the nuance in speech, etc.
Human brains differ from animal brains mainly in the way we can figure out hidden causes, intentions and motivations.
Intelligence is demonstrated based on whether we can successfully anticipate others' behavior and create a win-win situation.
Understanding human behavior, personality, needs and values is the key to better serving customers and making recommendations.
We provide tools for better prediction on multi-modality time series that might be affected by various factors.