Introducing QoreTune: Optimize Your Edge AI with Reservoir Computing for Enhanced Anomaly Detection
QuantumCore, headquartered in Shinagawa, Tokyo, has announced the launch of
QoreTune, an innovative parameter tuning tool designed specifically to enhance the performance of its vibration anomaly detection solution,
VADQore. This tool leverages the company's advanced reservoir computing technology, enabling users to optimize detection parameters intuitively through a user-friendly interface. In just ten seconds of normal data, QoreTune can create high-accuracy anomaly detection models.
Key Features and Functionality of QoreTune
QoreTune's primary function is to streamline the workflow from data collection to analysis and optimization. The tool operates seamlessly with the
VADQore View tablet dashboard, which is already in the market, allowing for comprehensive management of devices through Bluetooth connectivity. This integration ensures a smooth, offline workflow that supports industries where cloud connectivity may not always be feasible.
A Comprehensive Edge AI Solution
The VADQore series has stood out for its capacity to detect anomalies in real time using vibrational and acoustic data derived from minimal sample sizes of normal data. The recently launched QoreTune is dedicated to fine-tuning the VADQore's detection capabilities. It optimizes threshold levels and parameters tailored to diverse industrial devices and environments, making it significantly easier for users without specialized knowledge to operate the system.
Optimization Cycle: Data Collection, Analysis, and Application
1.
Data Collection: Users can connect VADQore View with VADQore devices, visualizing real-time vibrational and acoustic data. Additionally, data from multiple devices can be exported in CSV format.
2.
Data Analysis & Optimization: Data from VADQore View can be imported into QoreTune, which helps in optimizing reservoir computing parameters by visualizing anomaly score distributions and detection accuracies.
3.
Configuration Application: Once optimal settings are determined, these configurations can be directly applied to VADQore devices, enhancing detection accuracy.
By continuously refining these cycles based on operational data, organizations can significantly bolster their anomaly detection accuracy while utilizing an offline workflow that circumvents cloud dependency.
Limited Launch Demonstrations
QuantumCore is actively showcasing QoreTune along with its related solutions at the ongoing
AI Expo 2025 (Spring), held at Tokyo Big Sight from April 15-17, 2025. Attendees are encouraged to visit Booth AI-42 in the AI and Deep Learning Zone to experience firsthand the efficiency of QoreTune in improving detection capabilities.
Special Features of QoreTune
- - Data Import & Preview: Users can easily import normal (OK) and abnormal (NG) dataset files, streamlining the analytics process.
- - Intuitive UI for Parameter Settings: Specialized parameters, such as internal unit counts and spectral radii, can be adjusted visually, making the system user-friendly for non-technical users.
- - Preprocessing Settings for Signal Types: Optimize parameters specifically for signal types like audio and vibration to ensure the correct application in various environments.
- - Comprehensive Analysis Display: Visualization and calculation of precision, recall, F1 scores, and ROC/PR curves are provided, allowing for intuitive assessment of configuration effectiveness.
- - File Output and Application: Optimize configurations can be automatically generated and applied directly to VADQore devices, inclusive of an HTML report for effective team communication.
Future Perspectives
QuantumCore is committed to further developing the Qore series, expanding its applications beyond vibration and acoustic anomaly detection into other domains. The emphasis will be on preventive maintenance and quality management across manufacturing, construction, and logistics. The company aims to strengthen the functional integration between VADQore View and QoreTune, optimizing the use cycle for enhanced efficiency. By focusing on offline anomaly detection pipelines, the company looks to minimize cloud dependence while ensuring security and privacy.
Founded with guidance from leading researchers at the University of Tokyo and other institutions, QuantumCore is dedicated to providing flexible, real-time learning solutions for processing multivariate time series data with minimal data requirements. The company's vision is to unlock new possibilities in domains where data collection and computational resources are constrained.