Lizzie Coles-Kemp, Royal Holloway University of London: Privacy Awareness: An...
CN presentation
1. Security and Privacy
Implications of Pervasive
Memory Augmentation
Authors: Nigel Davis, Adrian Friday, Sarah Clinch, Corina Sas,
Marc Langheinrich, Geoff Ward, Albrecht Schmidt.
Presented By: Aadil Ahmed Adam
Sindhuja G Naik
2. Contents
• Introduction
• Related work in memory augmentation
• Application domains
• Architecture of present system
• Future architecture system
• Security and privacy threats
4. Continued…..
• New ways of augmenting human memory:
Near-continuous collection of memory cues.
Advances in data storage and processing
enables mining of stored cues for proactive
presentation.
Presence of ubiquitous displays.
• Capturing large amount of data of an
individual’s experiences and using it to trigger
recall is useful but raises many security
concerns.
5. Related work in memory augmentation
• Capturing of digital data for total recall, e.g. SenseCam photos.
• Questions against lifelogging’s total recall?
• Autobiographical memories are resilient to deliberate efforts to forget.
• Disposal of digital artifacts through selective deletion is suggested.
• Future work on forgetting needed to preserve privacy in lifelogging
technologies.
6. Application domains
1. Behavior change
• Important objective in healthcare,
transportation etc.
• Difficulty in implementing planned
behavior.
• Intentional behaviors are likely to be
implemented when individuals are
reminded of their own attitude.
• Realistic scheduling is important.
• Behavior is performed if its perceived
as achievable and enjoyable.
7. Continued….
2. Learning
• Use of ambient displays, reinforcing the learning of a wide range of skills.
• E.g. A study-abroad student could learn culturally-significant facts as they
explore a new city.
8. Continued….
3. Supporting failing memories
• Un-cued recall vulnerable to age-related
decline.
• Provide older people with time-reverent
and context-appropriate cues.
• Enjoy greater self-confidence and better
relationships.
4. Selective recall
• Attenuate the spontaneous retrieval of
related but un-reviewed memories.
9. Continued….
5. Advertising
• Users have memories triggered explicitly to
drive purchasing decisions.
6. Social acceptance
• Image based lifelogging devices not accepted.
• Location information tracked by mobile
devices improve memory construction.
• Non image based lifelogging devices provide
wealth of information.
• Cultural differences exist in terms of
technology acceptance.
10. Architecture of present system
• Experience data was gathered by devices worn or carried by a user.
• Stored or uploaded to cloud-based servers.
11. Continued…..
• The visual and auditory channels are dominant and recording these are a
focus in humans.
• Capturing meta information like time and location adds significant value
to the data.
Disadvantage:
• Relies on data captured exclusively by a specific user.
• Reduces the number of data streams available hence reducing the quality.
• E.g. using microphone on a mobile device in user’s pocket offers poorer
results than using a high-quality audio conferencing microphone built into
meeting room.
12. Architecture of future system
• Rely on the ability to appropriate screen real-estate from the large number
of displays that the user already looks at as part of their daily activities. E.g.
Photo frames, google mail
• Forms complex eco-systems of experience capture, storage and
presentation devices rather than the user-centric approaches employed.
13. Security and privacy threats
1. Experience provenance
• Data streams that constitute an individual’s memories are sourced from
devices not worn by the user.
• These data sources represent a point of attack against pervasive memory
augmentation.
• The challenge is that user may review captured experience long after the event
and it is impossible to detect that the original data stream was defective.
• Necessary to develop solutions that are able to provide end-to-end guarantees
for the user of the provenance of the data.
14. Continued…
2. Memory Protection
• Data store is highly distributed and will be accessed by a wide range of
parties.
• Applications designed to support recall may require access to this data.
• Will require sophisticated access control mechanisms with simple user
interactions.
• Managing digital assets after death is starting to attract research attention.
• Inheritance, ownership, and control issues pose challenge with regard to
the range of stakeholders involved.
• In developing solutions to these challenges, researchers need to reassure
users that their memories will be protected not just for short term but for
years to come.
15. Continued….
3. Memory manipulation
• Cued recall can be both used to re-enforce and attenuate human
memories.
• E.g. advertises and brand companies could pursue campaigns to forget
bad experiences and “only remember the good times”.
• Cues and memory need not be generic but can be specific to each
individual leading to more effective forms of memory manipulation.
• Key challenge – how can user tell if their memories are being
manipulated?
• Necessary to enable users to instantiate real-time monitoring of the cues.
• E.g. virus checker for our memories monitoring activity to identify
suspicious patterns.
16. Continued….
4. Privacy of bystanders
• Use of personal capture technology would impact the privacy of
bystanders.
• Challenge- how to protect bystanders while allowing substantial data
collection for human memory augmentation?
• Focus on technologies that do not actually record anything but instead
work like simple detectors. E.g. audio detectors.
• In video recording devices, instead of high-fidelity video capture, only
certain abstract elements of scene is recorded.
• Solutions are likely to combine elements of new technologies for creating
abstract recordings with a robust way of announcing recording practices
and policies to users.
17. Conclusion
• While benefits of pervasive memory augmentation are significant, such
systems challenge areas of security and privacy.
• Despite increasing ability to produce and store information, our society
still follows approach of selective capture and storage.
• Once memory augmentation systems become mainstream, we may see
radical transition from selective preservation to preserve everything and
only selectively remove parts which are inappropriate.