With these research projects we hope to provide the computational and communication foundations for sustainable development in core environmental, economic, and social sustainability areas

Energy-constrained and sustainable deep learning

The estimation and optimization of energy consumption and carbon footprint derived from the use of AI systems. In this line, we will foster change in the entire lifecycle of AI products (i.e., idea generation, training, re-tuning, inference, etc.) by characterizing the use of IT towards greater ecological integrity. The final goal is to develop AI technologies (particularly deep learning) that are compatible with sustaining environmental resources by exploring the trade-offs of accuracy of AI outcomes vs. reducing carbon emissions and computing power.



AI-Driven Classifier Building Pipeline

Federated Machine Learning

Learning-based Dimensionality Reduction

Meditron

ML-Enabled IoT Devices and Embedded AI

MLbench

Multi-Objective Machine-Learning Based Resource Management

Power-Aware Acceleration of Deep Learning (DL)

Training GANs: A Convex Optimization Perspective

Energy-constrained trustworthy systems

The creation of cryptocurrency systems that not only does not consume a lot of energy but can help save energy. The target is to create a new generation of cryptocurrency systems that can reward users who plant trees, and the underlying infrastructure does not use expensive consensus protocols.

Automated bug-hunting

Code Sanitization for Vulnerability Pruning and Exploitation Mitigation

Controlling Information Diffusion in Gossip Protocols

DICER

Enhancing Security of FPGAs in the Cloud

Full-System API Inference to Enforce Security

IoT-D

MultiSan

Performance Contracts for Software Network Functions

PowerSGD

PriBots

PrivySeal

SecureCells

Stainless

Transactional Computing

Verifying Software Network Functions with No Verification Expertise

VigNAT

Sustainable smart cities and transport systems

The application of IT and cloud computing technologies to support the creation of sustainable digital twins, including IoT data collection, to model cities and transportation systems. The target is to bring IT as an enabler of a progressive systemic change towards developing an entire value chain to address circularity and systematic carbon footprint reduction.



Armasuisse Project

Embedded AI for Aerospatial Navigation

Framing FFs

Urban Structure and Dynamics

UrbanTwin

Scientific computing

The creation of sustainable IT approaches and strategies to support the required massive computational and storage resources of different strategic scientific domains for our society with cloud computing. In particular, we will consider climate simulation, particles collider, and the Square Kilometer Array (SKA) challenges in the ETH domain.



Genomic and Reconfigurable Computing

Time-data Trade-off