Recognition Markets and Visual Privacy
1 like570 views
The document discusses recognition markets and visual privacy, focusing on the socio-technical systems that enable image recognition and the ethical considerations that arise from their use. It examines the implications for visual privacy, the design of recognition systems, and emphasizes the need for responsible implementation and education of participants. The impact of technological limitations and the roles of image aggregators in managing recognition spam are also highlighted.
![Thank you Ryan Shaw UC Berkeley School of Information [email_address] http://aeshin.org/](https://image.slidesharecdn.com/recognition-markets-and-visual-privacy-20545/85/Recognition-Markets-and-Visual-Privacy-25-320.jpg)
Recognition Markets and Visual Privacy
- 1. Recognition Markets and Visual Privacy Ryan Shaw UnBlinking Symposium UC Berkeley School of Information November 4, 2006
- 2. Photo by Chance Heath, http://flickr.com/photos/83664946@N00/132100361/
- 3. From the “420 Photo Album” at the CU-Boulder Police Department website
- 4. From the “420 Photo Album” at the CU-Boulder Police Department website
- 5. Some definitions Recognition The linking of images to the people, places, things or concept which they depict Recognition markets Socio-technical systems which use “human-in-the-loop” approaches to solve image recognition problems, often modeled as economic systems in which recognition services are exchanged for some capital
- 6. Recognition markets and visual privacy High-level overview of recognition systems “ Socio-technical turn” in recognition system design Possible applications and implications for visual privacy Guidelines for ethical design & implementation
- 7. Signal-based recognition systems Images are collected and analyzed as collections of signal features These features are used to build a statistical model of the thing to be recognized A new image is provided to the model as input The model provides an answer as to whether the new image depicts that thing
- 8. excerpt from LSCOM Lexicon Definitions and Annotations Version 1.0
- 9. Peer production requirements (Benkler) Modularity for asynchronous collaboration Granularity for lightweight contributions Tools and processes for quality control Yochai Benkler, “Coase's Penguin,” Yale Law Journal , 2002
- 13. Forms of semiotic labor (Warner) Syntactic labor Formal logic, keyword indexing, automated feature extraction Semantic labor High-level description, summarization, synthesis, creation of classification structures Julian Warner, “Forms of labour in information systems,” Information Research , 2002
- 17. Massive Multi-player Border Watch “ I look at this as no different than the neighborhood watches that we've had in our communities for years and years.” Texas Governor Rick Perry “ We have to make sure these things are secure and that they can handle the tremendous amount of hits we know we'll have through the servers. You're going to get millions of people on it.” Perry campaign spokesman “ Eyes of Texas to be on border,” Houston Chronicle , June 2, 2006
- 20. Ethical design and implementation Enforcing contextually appropriate use of metadata Tracking provenance to combat recognition spam Understanding the limits of technological protections Defining the responsibilities of image aggregators Educating participants in recognition markets
- 21. Spam
- 22. Limits of technological protections Alexander & Smith, “Engineering Privacy in Public: Confounding Face Recognition,” PET 2003
- 23. Ethical design and implementation Enforcing contextually appropriate use of metadata Tracking provenance to combat recognition spam Understanding the limits of technological protections Defining the responsibilities of image aggregators Educating participants in recognition markets
- 24. “ We’ll have the capability, and it will be up to society to determine how we use it. Society will determine that, not scientists.” Rick Rashid, Senior VP, Microsoft Research “ Computing, 2016: What Won’t Be Possible?” New York Times , October 31, 2006
- 25. Thank you Ryan Shaw UC Berkeley School of Information [email_address] http://aeshin.org/