Differential Privacy for Multi-armed Bandits: What Is It and What Is Its Cost?
Based on differential privacy (DP) framework, we introduce and unify privacy definitions for the multi-armed bandit algorithms. We represent the framework with a unified graphical model and use it to connect privacy definitions. We derive and contrast lower bounds on the regret of bandit algorithms...
Gespeichert in:
| Hauptverfasser: | , , |
|---|---|
| Format: | Artikel |
| Sprache: | eng |
| Schlagworte: | |
| Online-Zugang: | Volltext bestellen |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| container_end_page | |
|---|---|
| container_issue | |
| container_start_page | |
| container_title | |
| container_volume | |
| creator | Basu, Debabrota Dimitrakakis, Christos Tossou, Aristide |
| description | Based on differential privacy (DP) framework, we introduce and unify privacy
definitions for the multi-armed bandit algorithms. We represent the framework
with a unified graphical model and use it to connect privacy definitions. We
derive and contrast lower bounds on the regret of bandit algorithms satisfying
these definitions. We leverage a unified proving technique to achieve all the
lower bounds. We show that for all of them, the learner's regret is increased
by a multiplicative factor dependent on the privacy level $\epsilon$. We
observe that the dependency is weaker when we do not require local differential
privacy for the rewards. |
| doi_str_mv | 10.48550/arxiv.1905.12298 |
| format | Article |
| fullrecord | <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_1905_12298</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>1905_12298</sourcerecordid><originalsourceid>FETCH-LOGICAL-a678-efc23f9623cca5767012895a7e58ad7c60862888045d3ee3a6eebcc5d25a3b293</originalsourceid><addsrcrecordid>eNpNz8tOAjEYBeBuXBj0AVzRF5ix09KbG6LjhUkQXZC4nPy0f0OTAUxbiby9CixcnZyzOMlHyE3D6omRkt1C-o77urFM1g3n1lySxWMMARNuS4SBvqe4B3egYZfo69dQYgVpg54-wNbHku_oxxoK7TLtCv2d_tVM210u0ytyEWDIeH3OEVk-Py3bWTV_e-na-3kFSpsKg-MiWMWFcyC10qzhxkrQKA147RQzihtj2ER6gShAIa6ck55LECtuxYiMT7dHUP-Z4gbSof-D9UeY-AGZUkc_</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Differential Privacy for Multi-armed Bandits: What Is It and What Is Its Cost?</title><source>arXiv.org</source><creator>Basu, Debabrota ; Dimitrakakis, Christos ; Tossou, Aristide</creator><creatorcontrib>Basu, Debabrota ; Dimitrakakis, Christos ; Tossou, Aristide</creatorcontrib><description>Based on differential privacy (DP) framework, we introduce and unify privacy
definitions for the multi-armed bandit algorithms. We represent the framework
with a unified graphical model and use it to connect privacy definitions. We
derive and contrast lower bounds on the regret of bandit algorithms satisfying
these definitions. We leverage a unified proving technique to achieve all the
lower bounds. We show that for all of them, the learner's regret is increased
by a multiplicative factor dependent on the privacy level $\epsilon$. We
observe that the dependency is weaker when we do not require local differential
privacy for the rewards.</description><identifier>DOI: 10.48550/arxiv.1905.12298</identifier><language>eng</language><subject>Computer Science - Learning ; Statistics - Machine Learning</subject><creationdate>2019-05</creationdate><rights>http://arxiv.org/licenses/nonexclusive-distrib/1.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,780,885</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/1905.12298$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.1905.12298$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Basu, Debabrota</creatorcontrib><creatorcontrib>Dimitrakakis, Christos</creatorcontrib><creatorcontrib>Tossou, Aristide</creatorcontrib><title>Differential Privacy for Multi-armed Bandits: What Is It and What Is Its Cost?</title><description>Based on differential privacy (DP) framework, we introduce and unify privacy
definitions for the multi-armed bandit algorithms. We represent the framework
with a unified graphical model and use it to connect privacy definitions. We
derive and contrast lower bounds on the regret of bandit algorithms satisfying
these definitions. We leverage a unified proving technique to achieve all the
lower bounds. We show that for all of them, the learner's regret is increased
by a multiplicative factor dependent on the privacy level $\epsilon$. We
observe that the dependency is weaker when we do not require local differential
privacy for the rewards.</description><subject>Computer Science - Learning</subject><subject>Statistics - Machine Learning</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2019</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNpNz8tOAjEYBeBuXBj0AVzRF5ix09KbG6LjhUkQXZC4nPy0f0OTAUxbiby9CixcnZyzOMlHyE3D6omRkt1C-o77urFM1g3n1lySxWMMARNuS4SBvqe4B3egYZfo69dQYgVpg54-wNbHku_oxxoK7TLtCv2d_tVM210u0ytyEWDIeH3OEVk-Py3bWTV_e-na-3kFSpsKg-MiWMWFcyC10qzhxkrQKA147RQzihtj2ER6gShAIa6ck55LECtuxYiMT7dHUP-Z4gbSof-D9UeY-AGZUkc_</recordid><startdate>20190529</startdate><enddate>20190529</enddate><creator>Basu, Debabrota</creator><creator>Dimitrakakis, Christos</creator><creator>Tossou, Aristide</creator><scope>AKY</scope><scope>EPD</scope><scope>GOX</scope></search><sort><creationdate>20190529</creationdate><title>Differential Privacy for Multi-armed Bandits: What Is It and What Is Its Cost?</title><author>Basu, Debabrota ; Dimitrakakis, Christos ; Tossou, Aristide</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a678-efc23f9623cca5767012895a7e58ad7c60862888045d3ee3a6eebcc5d25a3b293</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2019</creationdate><topic>Computer Science - Learning</topic><topic>Statistics - Machine Learning</topic><toplevel>online_resources</toplevel><creatorcontrib>Basu, Debabrota</creatorcontrib><creatorcontrib>Dimitrakakis, Christos</creatorcontrib><creatorcontrib>Tossou, Aristide</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv Statistics</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Basu, Debabrota</au><au>Dimitrakakis, Christos</au><au>Tossou, Aristide</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Differential Privacy for Multi-armed Bandits: What Is It and What Is Its Cost?</atitle><date>2019-05-29</date><risdate>2019</risdate><abstract>Based on differential privacy (DP) framework, we introduce and unify privacy
definitions for the multi-armed bandit algorithms. We represent the framework
with a unified graphical model and use it to connect privacy definitions. We
derive and contrast lower bounds on the regret of bandit algorithms satisfying
these definitions. We leverage a unified proving technique to achieve all the
lower bounds. We show that for all of them, the learner's regret is increased
by a multiplicative factor dependent on the privacy level $\epsilon$. We
observe that the dependency is weaker when we do not require local differential
privacy for the rewards.</abstract><doi>10.48550/arxiv.1905.12298</doi><oa>free_for_read</oa></addata></record> |
| fulltext | fulltext_linktorsrc |
| identifier | DOI: 10.48550/arxiv.1905.12298 |
| ispartof | |
| issn | |
| language | eng |
| recordid | cdi_arxiv_primary_1905_12298 |
| source | arXiv.org |
| subjects | Computer Science - Learning Statistics - Machine Learning |
| title | Differential Privacy for Multi-armed Bandits: What Is It and What Is Its Cost? |
| url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-06T15%3A27%3A14IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-arxiv_GOX&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Differential%20Privacy%20for%20Multi-armed%20Bandits:%20What%20Is%20It%20and%20What%20Is%20Its%20Cost?&rft.au=Basu,%20Debabrota&rft.date=2019-05-29&rft_id=info:doi/10.48550/arxiv.1905.12298&rft_dat=%3Carxiv_GOX%3E1905_12298%3C/arxiv_GOX%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true |