Matches in SemOpenAlex for { <https://semopenalex.org/work/W4384345651> ?p ?o ?g. }
- W4384345651 abstract "Fairness of machine learning (ML) software has become a major concern in the recent past. Although recent research on testing and improving fairness have demonstrated impact on real-world software, providing fairness guarantee in practice is still lacking. Certification of ML models is challenging because of the complex decision-making process of the models. In this paper, we proposed Fairify, an SMT-based approach to verify individual fairness property in neural network (NN) models. Individual fairness ensures that any two similar individuals get similar treatment irrespective of their protected attributes e.g., race, sex, age. Verifying this fairness property is hard because of the global checking and non-linear computation nodes in NN. We proposed sound approach to make individual fairness verification tractable for the developers. The key idea is that many neurons in the NN always remain inactive when a smaller part of the input domain is considered. So, Fairify leverages white-box access to the models in production and then apply formal analysis based pruning. Our approach adopts input partitioning and then prunes the NN for each partition to provide fairness certification or counterexample. We leveraged interval arithmetic and activation heuristic of the neurons to perform the pruning as necessary. We evaluated Fairify on 25 real-world neural networks collected from four different sources, and demonstrated the effectiveness, scalability and performance over baseline and closely related work. Fairify is also configurable based on the domain and size of the NN. Our novel formulation of the problem can answer targeted verification queries with relaxations and counterexamples, which have practical implications." @default.
- W4384345651 created "2023-07-15" @default.
- W4384345651 creator A5059626072 @default.
- W4384345651 creator A5090690054 @default.
- W4384345651 date "2023-05-01" @default.
- W4384345651 modified "2023-10-17" @default.
- W4384345651 title "Fairify: Fairness Verification of Neural Networks" @default.
- W4384345651 cites W1480909796 @default.
- W4384345651 cites W1932198206 @default.
- W4384345651 cites W2014352947 @default.
- W4384345651 cites W2019411264 @default.
- W4384345651 cites W2097246321 @default.
- W4384345651 cites W2100960835 @default.
- W4384345651 cites W2165073069 @default.
- W4384345651 cites W2543296129 @default.
- W4384345651 cites W2594877703 @default.
- W4384345651 cites W2730550703 @default.
- W4384345651 cites W2754537581 @default.
- W4384345651 cites W2762833920 @default.
- W4384345651 cites W2794609696 @default.
- W4384345651 cites W2809701591 @default.
- W4384345651 cites W2892430965 @default.
- W4384345651 cites W2898868990 @default.
- W4384345651 cites W2900153411 @default.
- W4384345651 cites W2913771223 @default.
- W4384345651 cites W2957311447 @default.
- W4384345651 cites W2963054787 @default.
- W4384345651 cites W2963116854 @default.
- W4384345651 cites W2963327228 @default.
- W4384345651 cites W2963857521 @default.
- W4384345651 cites W2963913218 @default.
- W4384345651 cites W2964031043 @default.
- W4384345651 cites W2967682612 @default.
- W4384345651 cites W2980222512 @default.
- W4384345651 cites W2999818223 @default.
- W4384345651 cites W3002398329 @default.
- W4384345651 cites W3006235430 @default.
- W4384345651 cites W3032152562 @default.
- W4384345651 cites W3101094684 @default.
- W4384345651 cites W3102139284 @default.
- W4384345651 cites W3103741452 @default.
- W4384345651 cites W3104896058 @default.
- W4384345651 cites W3108851260 @default.
- W4384345651 cites W3156188980 @default.
- W4384345651 cites W3160165418 @default.
- W4384345651 cites W3183369696 @default.
- W4384345651 cites W3193448347 @default.
- W4384345651 cites W3194157648 @default.
- W4384345651 cites W3194588521 @default.
- W4384345651 cites W3205478695 @default.
- W4384345651 cites W4284677512 @default.
- W4384345651 cites W4288359825 @default.
- W4384345651 cites W4289751798 @default.
- W4384345651 cites W4384302771 @default.
- W4384345651 doi "https://doi.org/10.1109/icse48619.2023.00134" @default.
- W4384345651 hasPublicationYear "2023" @default.
- W4384345651 type Work @default.
- W4384345651 citedByCount "0" @default.
- W4384345651 crossrefType "proceedings-article" @default.
- W4384345651 hasAuthorship W4384345651A5059626072 @default.
- W4384345651 hasAuthorship W4384345651A5090690054 @default.
- W4384345651 hasBestOaLocation W43843456512 @default.
- W4384345651 hasConcept C108010975 @default.
- W4384345651 hasConcept C111498074 @default.
- W4384345651 hasConcept C114614502 @default.
- W4384345651 hasConcept C118615104 @default.
- W4384345651 hasConcept C119857082 @default.
- W4384345651 hasConcept C134306372 @default.
- W4384345651 hasConcept C154945302 @default.
- W4384345651 hasConcept C162838799 @default.
- W4384345651 hasConcept C173801870 @default.
- W4384345651 hasConcept C17744445 @default.
- W4384345651 hasConcept C199539241 @default.
- W4384345651 hasConcept C33923547 @default.
- W4384345651 hasConcept C36503486 @default.
- W4384345651 hasConcept C41008148 @default.
- W4384345651 hasConcept C42812 @default.
- W4384345651 hasConcept C46304622 @default.
- W4384345651 hasConcept C48044578 @default.
- W4384345651 hasConcept C50644808 @default.
- W4384345651 hasConcept C6557445 @default.
- W4384345651 hasConcept C77088390 @default.
- W4384345651 hasConcept C80444323 @default.
- W4384345651 hasConcept C86803240 @default.
- W4384345651 hasConceptScore W4384345651C108010975 @default.
- W4384345651 hasConceptScore W4384345651C111498074 @default.
- W4384345651 hasConceptScore W4384345651C114614502 @default.
- W4384345651 hasConceptScore W4384345651C118615104 @default.
- W4384345651 hasConceptScore W4384345651C119857082 @default.
- W4384345651 hasConceptScore W4384345651C134306372 @default.
- W4384345651 hasConceptScore W4384345651C154945302 @default.
- W4384345651 hasConceptScore W4384345651C162838799 @default.
- W4384345651 hasConceptScore W4384345651C173801870 @default.
- W4384345651 hasConceptScore W4384345651C17744445 @default.
- W4384345651 hasConceptScore W4384345651C199539241 @default.
- W4384345651 hasConceptScore W4384345651C33923547 @default.
- W4384345651 hasConceptScore W4384345651C36503486 @default.
- W4384345651 hasConceptScore W4384345651C41008148 @default.
- W4384345651 hasConceptScore W4384345651C42812 @default.
- W4384345651 hasConceptScore W4384345651C46304622 @default.