Unparalleled Insights
from Sensitive Data

DQ0 guarantees the protection of your data through Differential Privacy. You maintain complete control over your data and anybody can use it for advanced analytics and machine learning without disclosing any private information.
 

Data Protection made in Europe

How can sensitive data be used safely? DQ0 is a platform for privacy-preserving analysis and machine learning. Get the most out of your data without revealing any secret information. DQ0 guarantees mathematically verifiable data protection. Use DQ0 to halt data leaks in your company, to create data marketplaces where external analysts can work safely with your data, or to use external data sources safely without having to trust another entity. With DQ0, you are fully compliant with all data protection regulations at all times. read more

For Data Officers

DQ0 is available as private cloud SaaS or on-premise, installed "in front of" your data storage solution. The data itself does not have to be changed for use in DQ0. The integration effort is minimal. Basically, all data sources can be connected. DQ0 works with structured data from SQL and No-SQL databases, file systems and cloud storage as well as with binary data such as images or measurement data.
read more

For Data Scientists

DQ0 was developed by data scientists for data scientists. In addition to the the core functionality of computing models on protected data, DQ0 offers features for model versioning and tracking of your experiments. You can talk to DQ0 with the freely available DQ0 CLI command line program, the integrated web application, or the DQ0 SDK, directly from your Jupyter development environment.
read more

Value Proposition

 

DQ0 for shared data science

Use the value of your data and create data marketplaces

The DQ0 data quarantine is a secure data enclave which makes it easy to develop data-driven solutions and smoothly manage access for external or internal teams. Sensitive data can be safely added to DQ0 and shared across teams or individuals. Internal and external teams can derive general information and predictive models from the datasets, while sensitive information is never at risk. Data processing and model execution is performed inside the data quarantine. Everything developed with DQ0 will remain under control of the data owning company or individual. DQ0 allows you to create data marketplaces and grant everyone access without worrying about privacy.

 

DQ0 for use of external data

Integrate external data sources the secure way

The DQ0 data quarantine provides a mathematically proven solution for privacy-preserving data science. And it does that without altering the data itself. Therefore, sensitive data sources can be safely used for modelling by internal or external analysts. With DQ0 there is no need for data transformation pipelines for anonymization etc. Datasets will be used “as is”. Operations on those datasets can use complete data sets, including sensitive information. Information obtained from the data sets is always checked by DQ0 according to the strict criterion of differential privacy. Therefore, external partners can also provide you with data without having to worry about the protection of their data.

 

DQ0 for information security

Save costs and reduce the complexity of your information security

With DQ0, cybersecurity risks can be reduced dramatically. Data officers only grant access to a selected few (the “data owner team”). Everybody else will access the datasets through the DQ0 platform (“data analysts”). Therefore, with one single solution the information security team can manage access to sensitive information across the whole organization. Employees with access to the data through DQ0 can never causes data breaches, because they were never in control of the sensitive information. DQ0 is a mathematically proven and certified platform for secure data access and can reduce business risks and IT management complexity.

 

DQ0 Quarantine

How does it work?

With DQ0, you get one-click privacy. The data does not move to the analysis; the analysis moves to the data. You do not need to implement expensive and error-prone data anonymization processes. Rather, with DQ0 data protection is implemented at the request level in a clever new way of differential privacy for machine learning. Fully compliant with GDPR.

The DQ0 Data Shield will be installed on premise right in front of your data warehouse solution or as a trusted cloud enclave with always-on end-to-end encryption. Either way the data is protected and at the same time securely available for advanced analytics for (external) data science teams.

DQ0 can handle any type of data: structured data as well as text documents, images or any other format.

The DQ0 SDK offers a standardized interface for generally available models as well as for self-developed models.

Frequently Asked Questions

Because DQ0 relies on the mathematically secure concept of Differential Privacy and at the same time offers a well thought-out platform for data science. DQ0 sets the highest standards for the protection of your data.
DQ0 sets the highest standards in data protection and data security. We have a legal opinion, which certifies that DQ0 is suitable for compliance with data protection legislation. Contact us at dq0@gradient0.com for more information.
Differential privacy makes data protection measurable. With every data query something is revealed about the data (otherwise the query itself would be meaningless). Questions to supposedly completely anonymized (or pseudonymized, or masked) data also reveal information about the original data. And usually much more than desired. Differential Privacy sets a mathematically defined standard for this unwanted information retrieval, and DQ0 guarantees that it remains as low as possible.
Because the process of producing the synthetic data must also be safe. Synthetic data promise the same properties as the original data set without disclosing sensitive information. Unfortunately, this promise is unsustainable. As of the current state of research, this is simply not possible, not even in the foreseeable future. Solutions that promise otherwise should be treated with great caution.
Yes. But if this data previously contained personal information and was anonymized or pseudonymized (or masked), it is not really anonymous, but allows conclusions to be drawn about secret data. There are innumerable cases that show, for example, how certain persons could be attributed to seemingly completely anonymized data (membership disclosure). "This is [..] why Cynthia Dwork […] likes to say "anonymised data isn’t" – either it isn’t really anonymous or so much of it has been removed that it is no longer data." From "The Ethical Algorithm" by Michael Kearns and Aaron Roth. (link)
Data protection is always associated with costs. To protect a person's personal data, noise is added. And that in turn affects the data analysis results. DQ0 ensures strict data protection guarantees with precisely calibrated noise in order to achieve the best utility-privacy trade off.
DQ0 implements all data protection methods carefully and securely and tests this security continuously according to the latest scientific criteria. DQ0 is encrypted end-to-end. The software to control the platform is certified by TÜV Austria as "Trusted Application". (link)
And yet few do. Differential Privacy is a mathematical concept, not a solution. The reliable implementation of this method requires the interaction of experienced experts from mathematics, data science and computer science as they come together at Gradient Zero. DQ0 offers a tested and verifiable solution for data protection in machine learning.
DQ0 ensures the safe implementation of this budget. The level of protection is always transparent to data owners. At the same time, DQ0 ensures maximum usability of the data, even for sophisticated data science analyzes.
Differential privacy can also be applied directly to the data. Ask our experts about individual DQ0 solutions e.g. for a decentralized application of differential privacy. Contact us: dq0@gradient0.com.
DQ stands for data quarantine, the secure enclave for your data. We borrowed the Zero from our company Gradient Zero.