The HAR.S.H. project (HARdware aware extreme-scale Similarity search), carried out by the Computer Science Department of the University of Crete, reinforces its leading role in advanced data science and artificial intelligence research.

Ordered sequences of data points, known as data series, are one of the most common types of data. They are present in virtually every scientific and social domain. They appear as audio sequences, shape and image data, financial, telecommunications, environmental monitoring and scientific data, and they have many diverse applications, e.g., in health care, earth sciences, astronomy, biology, economics, etc. The exponential growth of data generated by such applications in all of the above areas leads to the need for efficient analysis of large collections of high-dimensional data series. At the same time, the proliferation of machine learning has enabled performing demanding tasks, albeit on data of specific formats. However, modern applications require processing multimodal data, i.e., data that is not in the same format (e.g., text, images, video). The efficient processing of such data collections involves many research challenges but is imperative as a unified representation of the different data types will allow processing of information over a significantly wider range.

HARSH will provide innovative solutions in the following directions:  

• Encoding Multimodal data. HARSH will address the need to encode high-dimensional objects such as audio, images and videos, enabling data series to capture multi-dimensional data.
• Utilizing the full computational power of modern computing platforms in an agnostic way. HARSH will exploit all computing elements of modern computing platforms, including not only multiple nodes, but also the multi-core capabilities of each node, as well as the full capacity of the attached accelerators. 
• Hardware-Aware Algorithms and Data Structures for Data Series Processing. We will focus on emerging memory technologies and study how the utilization of such technology can influence (or add to) the foundations of data series processing.

HARSH will demonstrate its value proposition in the following ways: 

• Use case 1 – Similar Document/File Finding Application.
• Use Case 2 – Photo Analysis for Travel Profile Enhancement Application.
• Use case 3 – Public Opinion Analysis Application. 

The HAR.S.H. project (ΥΠ3ΤΑ-0560901) was launched in April 2025 and will be active until May 2026. It is part of the initiative “SUB1.1 – Research Excellence Partnerships (REP)”.

The project is implemented under the National Recovery and Resilience Plan “Greece 2.0”, with funding from the European Union – NextGenerationEU

MORE INFORMATION
harsh-project.eu
www.linkedin.com/company/108218606
www.facebook.com/harshproject