Applications
Focus: Real-world impact
One of our focus is to deploy novel and scalable AI systems for real-world impact in order to rigorously understand last-mile challenges in novel systems design. Application deployment is based upon our research in novel data systems and is guided by three principles:
- Novel system designs that can thrive with domain expertise
- Iterative feedback loop with real-world use cases
- Human well-being and privacy focused design
Komorebi: Ecosystem monitoring at scale
Collaborators: ETH (Crowther Lab, IRIS), MIT, WWF, Stanford
Project Website: external page climateai.org
Our planet and humanity is facing an unprecedented climate crisis. We need to act now.
Land-Use and Land-Use Change (LULUC) such as Deforestation and Agriculture is responsible for 25% of global emissions. DS3Lab's Climate+AI Initiative aims to empower frontline communities by pushing the limits of current land-use monitoring and providing accessible technology for everyone.
News
Kara: Privacy-preserving medical data markets
Collaborators: Oasis Labs, UC Berkeley, Stanford Medical School
Project Website: external page kara.cloud
Kara is a privacy-preserving tokenized data cloud for medical data. Medical data is currently locked in data silos due to regulations and policies.
Kara leverages a distributed ledger and smart contracts to transparently log all transactions on your data. Smart contracts are self-enforcing and make sure that data consumers play by your rules. Kara uses trusted hardware to guarantee integrity and confidentiality for your data. Differential privacy is applied on the application level to prevent data leakage.
We build a privacy-first data cloud on smart contracts, trusted hardware and diffential privacy.
News
external page WIRED, external page The New York Times, external page MIT Technology Review and ETH News
Piximi: Democratizing ML for cell biology
Collaborators: Broad Institute, FIMM
Project Website: external page piximi.gitbook.io
DS3Lab, together with Broad Institute of MIT and Harvard and FIMM Helsinki is pioneering a new next-generation deep learning classifier for cell biology. By democratizing machine learning via browser-based execution, we hope to allow scientist to use machine learning in a more efficient and easier-to-use way.