Extraction of Complex DNN Models: Real Threat or Boogeyman?

Tutkimustuotos: Artikkeli kirjassa/konferenssijulkaisussaConference contributionScientificvertaisarvioitu

1 Lataukset (Pure)

Abstrakti

Recently, machine learning (ML) has introduced advanced solutions to many domains. Since ML models provide business advantage to model owners, protecting intellectual property of ML models has emerged as an important consideration. Confidentiality of ML models can be protected by exposing them to clients only via prediction APIs. However, model extraction attacks can steal the functionality of ML models using the information leaked to clients through the results returned via the API. In this work, we question whether model extraction is a serious threat to complex, real-life ML models. We evaluate the current state-of-the-art model extraction attack (Knockoff nets) against complex models. We reproduce and confirm the results in the original paper. But we also show that the performance of this attack can be limited by several factors, including ML model architecture and the granularity of API response. Furthermore, we introduce a defense based on distinguishing queries used for Knockoff nets from benign queries. Despite the limitations of the Knockoff nets, we show that a more realistic adversary can effectively steal complex ML models and evade known defenses.
AlkuperäiskieliEnglanti
OtsikkoEngineering Dependable and Secure Machine Learning Systems
AlaotsikkoThird International Workshop, EDSMLS 2020, New York City, NY, USA, February 7, 2020, Revised Selected Papers
ToimittajatOnn Shehory, Eitan Farchi, Guy Barash
Sivut42-57
Sivumäärä15
ISBN (elektroninen)978-3-030-62144-5
DOI - pysyväislinkit
TilaJulkaistu - 7 marraskuuta 2020
OKM-julkaisutyyppiA4 Artikkeli konferenssijulkaisuussa
TapahtumaThe AAAI-20 Workshop on Engineering Dependable and Secure Machine Learning Systems - Hilton, Midtown, New York, Yhdysvallat
Kesto: 7 helmikuuta 20207 helmikuuta 2020
Konferenssinumero: 20
https://sites.google.com/view/edsmls2020/home

Julkaisusarja

Nimi Communications in Computer and Information Science
KustantajaSpringer
Vuosikerta1272
ISSN (painettu)1865-0929
ISSN (elektroninen)1865-0937

Workshop

WorkshopThe AAAI-20 Workshop on Engineering Dependable and Secure Machine Learning Systems
LyhennettäEDSMLS
MaaYhdysvallat
KaupunkiNew York
Ajanjakso07/02/202007/02/2020
www-osoite

Sormenjälki Sukella tutkimusaiheisiin 'Extraction of Complex DNN Models: Real Threat or Boogeyman?'. Ne muodostavat yhdessä ainutlaatuisen sormenjäljen.

Siteeraa tätä