DEPLOYING MPC DE FOR SOCIAL GOOD FO Lucy Q Lu Qin in Re Resear arch Soft ftwar are Engineer | | Soft ftwar are & Applicat ation Innovat ation Lab ab | | Boston Un University Andr Andrei i La Lapets, , Frederick Jan ansen, , Ki Kinan an Ba Bab, Peter Floc Flockha hart, Rawan Raw ane Is Issa, , Ma Mayank Va Varia, , Az Azer Be Bestavros os
MPC FOR SOCIAL GOOD Combatting Pay Equity Medical Data Opioid Addiction Sharing Detecting Serial Economic Social Science Perpetrators of Inclusion Research Sharing Sexual Misconduct
MPC FOR SOCIAL GOOD Combatting Pay Equity Medical Data Opioid Addiction Sharing Detecting Serial Economic Social Science Perpetrators of Inclusion Research Sharing Sexual Misconduct 3
SECURE MULTI-PARTY COMPUTATION (MPC) pri rivate inputs f(s 1 , s 2 , s 3 ) = z pu publ blic ou output
CLOSING THE WAGE GAP IN BOSTON 1. 1. Un Understan anding the root ca causes of th the wage gap 2. 2. Closi Cl sing th the gap 3. 3. Ev Evaluati ting su succe ccess ss Over 200 Businesses
ORIGINALLY PROPOSED WORKFLOW Salary Data Business A Business B Results! Trusted 3 rd Party Salary Data
ORIGINALLY PROPOSED WORKFLOW Salary Data Business A Business B Results! Trusted 3 rd Party Salary Data
ROLES & CONCERNS Lawyer BWWC HR/Diversity IT Personnel Personnel
ROLES & CONCERNS Lawyer BWWC HR/Diversity IT Personnel Lia Liabilit ility Personnel
ROLES & CONCERNS Lawyer BWWC HR/Diversity IT Personnel Lia Liabilit ility Pa Participation Personnel
ROLES & CONCERNS Lawyer BWWC HR/Diversity IT Personnel Lia Liabilit ility Pa Participation Personnel Us Usab ability
ROLES & CONCERNS Lawyer BWWC HR/Diversity IT Personnel Lia Liabilit ility Pa Participation Personnel Au Audit ditabilit ility Us Usab ability
MPC SOLUTION Con Contribu butor or A BU Server BU S An Analyst st at BW BWWC (w (web se server/database se) (client running web browser) random masked masked masked actual masked masked + = + = mask data data data aggregate data aggregate A A A B data A data _ Analyst can never access this data random masked actual + = mask data data random random + = aggregate B B B mask mask mask random A B mask A = random mask B actual aggregate Public-key Encrypted Storage data only Analyst has key; no one else (including BU) Contribu Con butor or B can read the content of this data
100talent.org
ERROR CHECKING 15
USER EXPERIENCE & WEB ANALYTICS
USER EXPERIENCE & WEB ANALYTICS
MPC ENABLED FOR DATA ANALYSIS 2016 2017 total # of employers 69 114 # employees (1000s) 113 167 % of workforce 11 16 total annual earnings $11b $15b
LONGITUDINAL DATA
IMPROVED DATA GRANULARITY
FROM DATA TO IMPACT
SUPPORTING LOCAL MINORITY-OWNED BUSINESSES
DETECTING SERIAL PERPETRATORS OF SEXUAL MISCONDUCT 6x 3x 15% more likely to seek out more likely to report of f su surv rvivors rs were re matched me medical and emo motional thro th rough a sch school’s s wi with another survi vivo vor wh who support su rt services aft after Callisto Ca sto we webs bsite than to reporte re rted th the sa same using Ca us Callisto sto th their r sch school or r th the police ce perpet per petrator
Identifying information about a survivor and the accused can only be decrypted by a lawyer when at least 2 users name the same perpetrator
DETECTING SERIAL PERPETRATORS OF SEXUAL MISCONDUCT Alice Accused: Mallory Mallory Alice Accused: Mallory Bob Mallory Bob Accused: Eve Carlos Eve Carlos
DETECTING SERIAL PERPETRATORS OF SEXUAL MISCONDUCT share sl slop ope value (y) Bob y = ax + k Alice Carlos ha hashe hed d ke key re repre rese sented us user er id id as y-in as inter ercep ept where (x, y) is a share hashed user id (x)
/multiparty/umbral /multiparty/oprf 27
DEPLOYING MPC FOR SOCIAL GOOD
THANK YOU @lc lcyqn multiparty.org mu /m /multiparty @bu bu_compu putin ing sai sail.bu.edu /hi hics csail @hi hics csail 29
Recommend
More recommend
