Comments (2)
Hey @k-o-ta , can you explain how you would propose the agg service:
- Scales different contributions with different values of
scalingFactor
? - Applies noise when different contributions have different values of
scalingFactor
?
I am concerned this will be challenging to do privately. One of the properties of the existing system is that you can do rescaling after applying noise (and therefore is just post-processing private aggregates). It seems with your proposal you might need to do scaling before applying noise.
from attribution-reporting-api.
I apologize for the delay in my response. I understand your concerns.
- Scales different contributions with different values of scalingFactor?
Yes- Applies noise when different contributions have different values of scalingFactor?
Yes
As you stated, we would lose the characteristic of being able to rescale after applying noise.
I was considering rescaling after applying noise to aggregatable reports with the same scaling factor.
For example,
- three aggregatable reports with a scaling factor of 100. The values are 600, 500, and 400.
- In this case, the aggregation service calculates 100*(600+500+400)+Noise1, then rescales to get 600+500+400+Noise1/100.
- two aggregatable reports with a scaling factor of 200. The values are 300 and 200.
- In this case, the aggregation service calculates 200*(300+200)+Noise2, then rescales to get 300+200+Noise2/200.
What the user ultimately gets is 600+500+400+Noise1/100+ 300+200+Noise2/200.
If we had initially set the scaling factor to 100, we would have gotten 600+500+400+300+200+Noise/100.
To summarize about Noise:
- Noise1/100 + Noise2/200 == (Noise1/100 + 0.5Noise2/100)
- Noise/100
In other words, we would compare Noise1 + 0.5Noise2 and Noise. If the distributions of Noise1, Noise2, and Noise are the same, this proposal might lead to a decrease in the accuracy of the summary report.
from attribution-reporting-api.
Related Issues (20)
- Aggregation Key bits order HOT 6
- Scalling aggregation service HOT 1
- Configuration of aggregation service for security screening HOT 1
- Versioning of aggregatable reports HOT 3
- Flex event explainer doesn't describe how debug reports are affected
- Flexible event-level explainer doesn't describe how "Max event-level reports per attribution destination" is affected HOT 1
- "max event-level reports per attribution destination" check may be wrong with respect to report replacement HOT 1
- Storage limits are inconsistent with Clear-Site-Data integration HOT 1
- Source's "number of event-level reports" field is never initialized
- Randomized source response should be a list, not a set
- Noised sources with no reports are handled differently from noised ones with reports HOT 3
- Flexible event-level explainer doesn't describe how randomized response is affected
- Feedback on consolidating Coordinator Services
- s
- Consider only calling attributed reporting origin limit once per trigger
- Aggregatable flexible contribution filtering
- Allow max attributions per rate-limit window during event report replacement
- App-to-Web Click Source Registration HOT 4
- Header validator inconsistent with spec re size checks
- Incorrect Laplace noise formula in your documentation HOT 1
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from attribution-reporting-api.