HomeTechnologyTigerGraph launches Workbench for ML / AI modeling of graphical neural network

TigerGraph launches Workbench for ML / AI modeling of graphical neural network


TigerGraphmanufacturer a graphical analytics The platform for information scientists introduced its new generation of TigerGraph ML (Machine Learning) Workbench tool during the Graph & AI Summit today, which will allow analysts to significantly improve the accuracy of the ML model and shorten development cycles.

TigerGraph VP Victor Lee told VentureBeat that Workbench does this by using familiar tools, workflows and libraries in a single environment that connects directly to existing data pipelines and ML infrastructure.

The ML Workbench Jupyter is a Python-based development framework that allows data scientists to create in-depth AI models using data directly from the business. The graph has proven to be an effective ML more accurate forecasting power and requires less processing time than the conventional ML approach.

Conditional machine learning algorithms based on the study of systems with training kits to develop a trained model. This pre-designed model is used to classify or recognize a set of test data; it can usually take days or weeks to complete for a particular use case. Graphically based ML can sometimes take several minutes to create an algorithmic model.

The cost of ML is high, but so is the learning curve

“Graphics have been proven to accelerate and improve ML learning and performance, but the learning curve to use APIs (application programming interfaces) and libraries to do this has proven to be very difficult for many data scientists,” Lee told a media consultant. “Thus, we have created the ML Workbench to provide a new functional layer between data scientists and graphics machine learning APIs and libraries to facilitate data storage and management, data preparation and ML training.

“In fact, we found that early implementers used ML Workbench and TigerGraph to increase the accuracy of ML models by 10-50%,” he said.

The whole mindset of TigerGraph revolves around the definition of a human personality based on your interactions with others, Lee told VentureBeat.

“The same goes for graphs in data modeling, and that now applies to neural networks.” Li said. “Every node in the chart is interconnected like people. Graphs are great for querying sample matching algorithms. The workbench will help you apply machine learning based on the information inside the graph, but the real power comes with graphical neural networks that have regular graphs on the steroid.

“For example, our DGL (deep graphics library) has a Pytorch geometric extension that supports (Meta) graphical neural networks,” he said. “It’s a great feature, and it shows that we’re going where the information scientists are; we are not trying to teach them anything new. We use tools that they already know and are comfortable with, because we are trying to reduce the learning curve. ”

Fraud is optimal for forecast use cases

The ML Workbench allows organizations to identify advanced concepts in node-prediction applications, such as fraud and external prediction applications. product recommendations, Lee said. Lee said the ML Workbench allows AI / ML practitioners to study graphics-enhanced machines and explore graphical neural networks (GNNs) because it is fully integrated with TigerGraph’s database for processing / manipulating parallel graphics data.

The ML Workbench is designed to work with popular in-depth learning frameworks such as PyTorch, PyTorch Geometric, DGL and TensorFlow, and gives users the flexibility to choose the frame they are most familiar with. Lee said the ML Workbench is ready for plug-and-play for both Amazon SageMaker, Microsoft Azure ML and Google Vertex AI.

The ML Workbench is designed to work with enterprise-level data. Users can learn GNNs – even on very large schedules – with the following internal capabilities:

  • TigerGraph DBs distributed storage and mass parallel processing;
  • Graphically based sections create training / validation / test schedule data sets;
  • Graph-based grouping for GNN mini-batch training to increase performance and reduce HW requirements; and
  • Example of subclause support advanced GNN modeling techniques.

The ML Workbench is as a fully managed cloud service and is compatible with TigerGraph 3.2, which is available for local use. The ML Workbench, currently available as a preview, will generally go on sale in June 2022, Lee said.

TigerGaph competes with Neo4J, ArangoDB, MemGraph and several others in the graphical database space.

‘Million Dollar Challenge’ winners were selected

Presented by TigerGraph at the Graphics and AI Summit winners -nin Schedule for the entire Million Dollar Challenge – To award $ 1 million to projects that analyze and solve the biggest global social, economic, health and climate issues of the day, change the game, and work on a schedule.

The winning projectswas announced this week Graphics + AI Summit, was manually selected by a global panel of judges from more than 1,500 registrations in more than 100 countries. The Mental Health Hero claimed the Grand Prize of $ 250,000 for creating a program to provide more access and personalization to mental health treatment.

Source link