0.25
0.5
0.75
1.25
1.5
1.75
2
Privacy and Background Knowledge
Published on 2007-02-253924 Views
The digitization of our daily lives has led to an explosion in the collection of data by governments, corporations, and individuals. Protection of confidentiality of this data is of utmost importance.
Presentation
Privacy and Background Knowledge08:18
An Abundance of Data34:17:03
Driving Factors: A LARGE Hardware Revolution43:45:33
Driving Factors: A small Hardware Revolution53:33:51
Other Driving Factors60:56:05
picture70:39:02
picture78:58:14
picture88:34:10
Pulsars89:34:38
Pulsar Surveys103:36:33
Project Requirements106:07:27
Driving Factors: Analysis Capabilities112:19:19
And Even the Popular Press Caught On123:20:03
Concerns About Privacy128:57:08
The Setup141:04:33
Model I: Untrusted Data Collector145:47:36
Minimal Information Sharing154:18:00
Model II: Trusted Data Collector165:40:44
Disclosure Limitations181:43:35
Types of Disclosure191:48:40
Types of Disclosure199:05:17
Types of Disclosure207:35:56
Talk Outline214:19:07
Privacy Preserving Associations230:07:03
Problem Introduction233:33:46
Example250:26:02
The Itemset Lattice267:50:48
Frequent Itemsets274:25:54
Breath First Search: 1-Itemsets280:49:12
Breath First Search: 1-Itemsets283:48:00
Breath First Search: 2-Itemsets286:53:51
Breath First Search: 3-Itemsets291:47:20
Breadth First Search: Remarks294:42:18
Depth First Search (2)298:13:02
Depth First Search (3)298:44:25
Depth First Search (4)302:00:27
Talk Outline304:29:00
Our Model307:47:21
Our Model (Contd.)315:40:39
Our Model (Contd.)317:48:02
Privacy Preserving Associations318:47:42
Minimal Information Sharing324:48:39
Our Model (Contd.)327:39:12
Our Model (Contd.)329:45:58
Our Model: Another View331:03:57
The Problem340:11:31
The Randomized Response Model346:00:11
Another View: Two Questions354:32:07
Analysis361:09:27
Analysis (Contd.)365:17:43
Interval Privacy373:41:03
Interval Privacy: Quantifying Privacy379:30:21
Talk Outline392:19:31
Background Knowledge395:56:54
Example: {a, b, c}405:52:58
Example: {a, b, c}413:17:46
Example: {a, b, c}413:55:37
Example: {a, b, c}418:05:13
Example: {a, b, c}426:39:21
Privacy Breaches436:03:43
Simple Privacy Breaches447:17:05
Privacy Breaches: Goals463:02:03
α-to-β Privacy Breach473:17:47
α-to-β Privacy Breach481:56:38
α-to-β Privacy Breach485:07:11
α-to-β Privacy Breach491:10:43
α-to-β Privacy Breach494:24:36
Amplification Condition506:27:31
Amplification Condition509:53:02
Amplification Condition514:46:22
Amplification Condition521:05:13
Amplification Condition524:14:21
The Bound on α-to-β Breaches533:11:56
The Bound on α-to-β Breaches543:22:32
Amplification: Summary546:35:19
Talk Outline564:02:43
Definition of select-a-size595:12:59
Definition of select-a-size598:57:09
Definition of select-a-size601:34:15
Definition of select-a-size607:05:06
Support Recovery612:44:23
Support Recovery615:37:33
Support Recovery622:13:42
Support Recovery624:04:52
Support Recovery625:41:26
Support Recovery626:49:24
The Unbiased Estimators627:37:25
Apriori [AS94]630:00:27
The Modified Apriori637:01:21
Talk Outline641:05:28
Talk Outline644:16:07
Trusted Data Collector653:35:54
Disclosure Limitations656:57:39
Sample Microdata662:21:49
Removing SSN …680:07:56
Linkage Attacks685:04:30
Linkage Attacks (Contd.)695:10:15
Quasi-Identifiers and Sensitive Attributes709:01:15
K-Anonymity [Sweeney02]715:25:46
K-Anonymity727:30:08
K-Anonymity Through Generalization729:42:10
Example Microdata733:06:43
4-Anonymous Microdata735:39:28
K-Anonymity Algorithms744:21:10
Talk Outline746:23:12
Example Microdata748:47:55
4-Anonymous Microdata749:34:57
Homogeneity Attack750:35:40
Background Knowledge Attack762:36:42
Data Publishing Desiderata782:54:51
Privacy Definition (1)792:14:37
Privacy Definition (2)797:52:21
Bayes-Optimal Privacy– Drawbacks806:22:06
Towards A Practical Definition (1)807:44:25
Towards A Practical Definition (2)808:44:53
Ensuring Diversity809:39:52
3-Diverse Microdata817:09:33
L-Diversity Revisited824:41:56
L-Diversity: Summary830:29:16
What I Talked About832:06:17
What I Talked About (Only Useful Stuff)842:48:29
Open Problems850:17:17
Modeling Belief875:40:37
Thanks878:34:42
But Of Course We Have More Confidence Than Scott Adams …883:44:58
Questions?885:53:32