DApp of the Week #9 — NSFW (AI Trained Model Monetization)
Not Safe For Work is a commonly used app that filters out pictures that can’t be used for work. It can be used in situations such as when a user loads a new picture onto a web service, so the service can automatically validate whether it is appropriate to be shown to other users.
The application can use several machine learning models. For example, “Not safe for school” or even “Not safe for anyone” trained models enable the app to identify these kinds of pictures.
💡 Want to learn more about iExec? Check out iExec Academy!
iExec Academy aggregates all content related to the project. You’ll find articles, tech documentation, videos, interactive demos, and much more! Whether you are a beginner or an expert, a developer or crypto-enthusiast, you’ll find what you are looking for on iExec Academy!
For the iExec NSFW dapp, the requester (end-user) can actually choose which model he wants to use to filter his/her image. The default model is this standard model, historically maintained by Yahoo. The end-user workflow is pretty straight forward: you choose the dataset you’re going to use, you select a picture to be analyzed and run the app.
When you click on ‘Buy analysis’, you can see that, as an end-user, you pay 1 RLC directly to the dataset owner, and 5 RLC to the worker executing the task.
As an end-user, you never access to the whole dataset: you simply get the results of the in-enclave processing. This is an example, in its simplest form, of how secure ‘data renting’ works. Requesters can use valuable datasets without actually being able to inspect them.
In the end, users get their analysis result, while dataset owners, devs and workers all rewarded in RLC. My cat gets a rating of 0.019177 which means that it can be used safely and you can see all the details of this execution on the iExec Explorer:
Training a machine learning model is not a trivial task: a huge amount of data needs to be sorted and labeled, before being processed (for transfer learning, etc). The model needs to be refined before reaching an acceptable level of accuracy. All this requires a huge amount of resources, including time and money. iExec provides the opportunity for data owners to set their own conditions of use and price per execution (see how in the official documentation).
Users don’t have access to the dataset itself: they just enjoy the output results.
You’re invited to try out the app (you’ll need both ETH and RLC tokens) You can join and share your feedback in Slack, Gitter or Telegram. Feel free to subscribe to our newsletter to be the first to know.
iExec recently launched the much anticipated iExec V3. See below for more information on the new feature and adoption announcements, including the fact that iExec V4 (High-performance computing and GPU) will be coming much earlier than planned.
Connect with us!
Website • Blog • Slack • Telegram • Reddit • Twitter • Facebook • LinkedIn• Youtube • Github • Kakao • Instagram • Steemit • Katacoda • Docs
