Six new CSE-projects funded by the Swedish Research Council

​The Swedish Research Council will distribute a total of SEK 1.1 billion to projects within natural and engineering sciences. The grants are for the period up to 2024.

SYNTM: Synthesis of Teamwork Multi-Agent Systems

Yehia Abd Alrahman, postdoc in the Formal Methods division

Yehia Abd Alrahman
The increasing adoption of robots in industry creates a demand to enable robots to interact in order to form teamwork Multi-Agent Systems (MAS) – a set of collaborative agents that interact in a functional way and behave as if they were a single capable agent, pursuing a set of joint goals. Teamwork MAS are important because of workload sharing and the fact that the capabilities of a team exceed the ones of an individual. For engineers, this is a potential to solve complex problems by breaking them on simple collaborative agents. This also opens the door for scientists to explain grand questions about societies, economies, and the emergence of complexities. 
Correct-by-design techniques that ensure the safety and reliability of these systems are not that advanced. In fact, existing synthesis algorithms can only produce individual robots or co-existing ones that are functionally independent. Given the substantial progress in artificial intelligence and robotic hardware, the time is now ripe to unleash the development of teamwork MAS that collectively behave according to their design goals. The purpose of SYNTM is to develop a foundational framework that will enable the automatic production of correct Teamwork MAS from high-level descriptions of desired behaviour. The findings of SYNTM will comprise a significant development of current theories, that are still unable to handle this class of systems.
Funding: 4 year starting grant, total of 4 million SEK.

AccessTable: Accessible Collaboration around Configurable Displays

Morten Fjeld, professor in the Interaction Design division

Morten FjeldCollaborative decision-making (CDM) technologies are an effective way to allow people to work together on complex tasks. The topic records technical advancements especially in the area of actuated reconfigurable screens and how people can do multi-device interactions. However, as the sophistication of this technology increases, it becomes more difficult to ensure that people with disabilities (PwD) are guaranteed access to the tools of collaboration. 

In this project we will examine the accessibility of current and future CDM technologies for PwD and develop a comprehensive view of best practice for the development of inclusive CDM. Exploring the accessibility landscape of collaborative tabletop technology is an important piece of foundational research. Along with this we will examine the accessibility of related future research directions, building replicas and simulations of significant CDM innovations through a combination of prototyping, robotic trolleys, workspace integration of tabletop and tablet devices, to guide development of multi-display surfaces that can augment user capabilities. We will evaluate the complex accessibility issues that emerge and put together a comprehensive set of guidelines for accessible CDM usage.
Funding: 4 year project grant, total of 4 million SEK.

P4PIM: Principles of power-constrained HPC programming for PIM networks

Miquel Pericas, associate professor in the Computer Engineering division

Miquel PericasComputer design has traditionally focused on increasing processor speed and memory capacity. This has created a bottleneck between processor and memory, as latencies and bandwidth struggle to accommodate the needs of data-intensive applications. Processing-in-Memory (PIM), embodied by products such as the Hybrid Memory Cube, describe an approach to overcome this bottleneck. Combined with silicon interposers, it becomes feasible to build chips with high memory capacity, high bandwidth and low latency. However, this transition requires to adapt the programming model to this new generation of chips. The goal of the P4PIM project is to enable data-intensive HPC on PIM-based systems, by researching the necessary runtime technologies and proposing extensions to widely used programming models.

The project is organized into three phases that will (1) study how to partition data and collocate it with its opertators, (2) research how to manage parallelism to meet the power constraints of the PIM systems, and (3) study runtime and architecture codesign to enable novel organizations across the computer architecture stack. A breakthrough in methods for partitioning data and managing parallelism in scalable, parallel PIM-based systems is expected to have a high impact and be groundbreaking, even beyond HPC and PIM based systems, making P4PIM's scientific significance and industrial relevance high.
Funding: 4 year project grant, total of 4 million SEK.

Enabling Reactive Synthesis through Runtime Verification

Nir Piterman, associate professor in the Formal Methods division

Nir PitermanReactive synthesis – automatic production of plans from high level specification – is emerging as a viable tool for the development of robots. Reactive synthesis from temporal logic specification is now a well known technique in applications of robotics. However, to date, this type of synthesis has had little impact on other fields. Our view is that the restrictions on usage of synthesis both due to the capacity of tools and due to the kind of specifications that can be supported are a barrier to the further usage of synthesis. 

In this grant, we suggest to combine techniques of reactive synthesis with runtime verification. We plan to use two advantages of runtime verification to further promote reactive synthesis. First, the ability of runtime verification to express specifications that go much beyond discrete-time temporal logic and regular languages. Second, the efficiency of translating these specifications to monitors. By using runtime verification techniques to handle parts of the specifications we will enable more expressive specifications that will be handled more efficiently. Our hope is to increase the utility of reactive synthesis in robotics and elsewhere.

The project supports one PhD student.
Funding: 4 year project grant, total of 4 million SEK.

A Programming Framework for Differential Privacy with Accuracy Calculations

Alejandro Russo, professor in the Information Security division

Alejandro Russo
The growing availability of large scale personal data has put forward the challenge to cope with privacy as a major one. The EU GDPR legislation for data protection contributed to creating such awareness. Differential privacy (DP) is currently the gold standard which enables data analysts to mine useful information from private datasets while protecting the privacy of individuals. A standard way to achieve DP is by adding statistical noise to the result of data analyses. DP is compositional for privacy: there are basic building blocks and compositional properties which ensure that the privacy guarantees come from the guarantees of the building blocks. Unfortunately, reasoning about accuracy is less compositional. It is not surprising that many DP tools and programming frameworks lack support for reasoning about the accuracy of composed analyses.
This proposal presents techniques to enable reasoning about accuracy in a compositional manner. We propose the novel idea to apply information-flow control techniques – originally designed to track how data flows within systems – in order to internalize the use of probabilistic bounds when calculating the accuracy of composed data analyses.

With this project we exploit adding computation capabilities to the memory so that operations can be performed where the data is stored, thus eliminating the inefficient data transfers. We will develop a new computational model in which code move around instead of data, which is more inefficient. This will yield computer systems with substantially higher processing rates and lower energy consumption.
Funding: 4 year project grant, total of 4 million SEK.

Also approved

Automated testing of boundaries for quality of AI/ML models 

Robert Feldt, professor in the Software Engineering division
Funding: 4 year project grant, total of 2,18 million SEK.

Published: Fri 06 Nov 2020.