NEC orchestrating a brighter world
NEC Laboratories Europe

Research Groups
Intelligent Software Systems

  • Intelligent Software Systems
  • Graph Machine Learning


Intelligent Software Systems

The Intelligent Software Systems (ISS) group develops energy-efficient technology that makes computer systems faster and more sustainable. Our research encompases computer architectures and networking, supercomputing, parallel computing, physics and machine learning, which includes the discovery of new approaches for relational and geometrical data.. This diversity fosters creativity and spurs the development of ideas that advance the state of the art and enable future advancements for humankind.

Scientific Computing

Our group’s vision is to create a world in which scientific advancements are not impeded by engineering limitations. In scientific computing and computational sciences, the bottleneck is often the inability to design software that can efficiently implement scientists’ groundbreaking ideas.

Our goal is to enable scientists to write programs that tell computers what to do, not how to do it. To accomplish this we develop technology at the crossroads of domain-specific programming languages, software engineering, compiler and runtime systems for AI and parallel computing.

Domain experts have a limited understanding of platform complexities, and yet their programs are mostly executed as is on target platforms, which results in inefficient use of resources.
NEC is developing technology that will use the domain expert program as a formal computation for intent specification. This will automatically generate an optimized software implementation for the target platform that will extract the best performance from its underlying hardware. Using this approach, experts from different disciplines will be able to make the best use of new hardware advancements (see Figure 1 below).

 

In addition to relational learning, we research geometric deep learning to advance technologies in computational science. With this research, we model the interaction of physical entities in a certain range of space. Conventional numerical simulation requires huge computing resources. By replacing simulation with machine learning, and improving on existing approaches, we are reducing the number and duration of scientific tasks needed for research. For example, we recently developed a surrogate model to solve partial differential equations that drastically reduces the computation load and cost of machine learning. With a sufficient amount of data, the model can even predict future changes to a physical system with no knowledge of its partial differential equations.

New software frameworks for accelerated computing

The ISS group build software frameworks that enable scientists to leverage new computer architectures and processors efficiently. We achieve this by combining automated algorithm search-and-selection functions with software synthesis and compiler techniques. In close collaboration with our colleagues working in the larger NEC family, we then apply our results. Examples include accelerated research in personalized medicine for curing cancer and discovering broadly protective beta-coronavirus vaccines.

As our world grows increasingly interconnected, cybersecurity becomes a critical component in ensuring our systems maintain high reliability and integrity.

Advanced security to counter evolving threats

Within the ISS group, we develop advanced system-level techniques to improve the observability of systems and scale monitoring efficiently and sustainably. In parallel, we devise techniques to extract, represent, and assess information about cyber threats from a variety of sources. These sources include cyber threat intelligence, social media and traditional news sources.

Cyber threat intelligence provides millions of text-based reports every month. Security experts need to process each report to understand what is relevant for the systems they monitor. NEC is working to make this process scalable by:

  • Enabling semantic search in cyber threat intelligence’s structured and unstructured sources.
  • Making it more efficient to observe network and host infrastructure.

This will let security experts quickly extract the cyber threat intelligence they need to monitor their systems (see figure 2 below).

Combining these two approaches yields deep insight into the behavior of the systems we operate. It provides decision support for cybersecurity at scale and enables human operators and artificial systems to work together to defend against cyberattacks.


Graph Machine Learning

Machine learning is a data-driven science. It generates computer programs from data to perform tasks, without being explicitly programmed to do so. The Graph Machine Learning group researches machine learning approaches for relational and geometrical data using GraphAI technology developed by us at NEC Laboratories Europe. We focus on exploring new ways we can apply machine learning that challenges standard (deep) learning methods and advances society.

Advancing relational learning with GraphAI technology

Our team researches graph-structured, relational learning. Using GraphAI, we analyze relational or graph-structured data that represents the features of data points, their relationships and interdependencies.
GraphAI helps overcome the problem of missing data points in complex data dependencies, meeting the needs of emerging use cases.

We address various issues of using machine learning for size of errors in its own predictions. Our approach integrates probabilistic modeling with deep graph networking to robustly model the propagation of graph uncertainties.

Discovering new materials with materials informatics

Recent applications of GraphAI technology have focused on the materials science industry. A NEC study of related use cases shows how leveraging GraphAI’s relational learning and geometric deep learning methodologies will help achieve a digital transformation in materials science and the chemical industry. With relational learning, knowledge between scientific documents and a product developer’s requirements can be linked. Geometric learning augments simulations such as molecular dynamics and ab initio theory–based calculations.

Through our research activities, the Graph Machine Learning group explores the frontiers of machine learning applications, helping find solutions to some of society’s most pressing technical challenges. We conduct research in computer science and numerous industry domains, including public services, manufacturing, finance, biomedical, physics, and chemistry.

Top of this page