Dataset Information

A Genetic Attack Against Machine Learning Classifiers to Steal Biometric Actigraphy Profiles from Health Related Sensor Data.

ABSTRACT: In this work, we propose the use of a genetic-algorithm-based attack against machine learning classifiers with the aim of 'stealing' users' biometric actigraphy profiles from health related sensor data. The target classification model uses daily actigraphy patterns for user identification. The biometric profiles are modeled as what we call impersonator examples which are generated based solely on the predictions' confidence score by repeatedly querying the target classifier. We conducted experiments in a black-box setting on a public dataset that contains actigraphy profiles from 55 individuals. The data consists of daily motion patterns recorded with an actigraphy device. These patterns can be used as biometric profiles to identify each individual. Our attack was able to generate examples capable of impersonating a target user with a success rate of 94.5%. Furthermore, we found that the impersonator examples have high transferability to other classifiers trained with the same training set. We also show that the generated biometric profiles have a close resemblance to the ground truth profiles which can lead to sensitive data exposure, like revealing the time of the day an individual wakes-up and goes to bed.

SUBMITTER: Garcia-Ceja E

PROVIDER: S-EPMC7497442 | biostudies-literature | 2020 Sep

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

A Genetic Attack Against Machine Learning Classifiers to Steal Biometric Actigraphy Profiles from Health Related Sensor Data.

Garcia-Ceja Enrique E Morin Brice B Aguilar-Rivera Anton A Riegler Michael Alexander MA

Journal of medical systems 20200915 10

In this work, we propose the use of a genetic-algorithm-based attack against machine learning classifiers with the aim of 'stealing' users' biometric actigraphy profiles from health related sensor data. The target classification model uses daily actigraphy patterns for user identification. The biometric profiles are modeled as what we call impersonator examples which are generated based solely on the predictions' confidence score by repeatedly querying the target classifier. We conducted experim ...[more]

PMID: 32929615

Similar Datasets

Project description:BackgroundTwitter presents a valuable and relevant social media platform to study the prevalence of information and sentiment on vaping that may be useful for public health surveillance. Machine learning classifiers that identify vaping-relevant tweets and characterize sentiments in them can underpin a Twitter-based vaping surveillance system. Compared with traditional machine learning classifiers that are reliant on annotations that are expensive to obtain, deep learning classifiers offer the advantage of requiring fewer annotated tweets by leveraging the large numbers of readily available unannotated tweets.ObjectiveThis study aims to derive and evaluate traditional and deep learning classifiers that can identify tweets relevant to vaping, tweets of a commercial nature, and tweets with provape sentiments.MethodsWe continuously collected tweets that matched vaping-related keywords over 2 months from August 2018 to October 2018. From this data set of tweets, a set of 4000 tweets was selected, and each tweet was manually annotated for relevance (vape relevant or not), commercial nature (commercial or not), and sentiment (provape or not). Using the annotated data, we derived traditional classifiers that included logistic regression, random forest, linear support vector machine, and multinomial naive Bayes. In addition, using the annotated data set and a larger unannotated data set of tweets, we derived deep learning classifiers that included a convolutional neural network (CNN), long short-term memory (LSTM) network, LSTM-CNN network, and bidirectional LSTM (BiLSTM) network. The unannotated tweet data were used to derive word vectors that deep learning classifiers can leverage to improve performance.ResultsLSTM-CNN performed the best with the highest area under the receiver operating characteristic curve (AUC) of 0.96 (95% CI 0.93-0.98) for relevance, all deep learning classifiers including LSTM-CNN performed better than the traditional classifiers with an AUC of 0.99 (95% CI 0.98-0.99) for distinguishing commercial from noncommercial tweets, and BiLSTM performed the best with an AUC of 0.83 (95% CI 0.78-0.89) for provape sentiment. Overall, LSTM-CNN performed the best across all 3 classification tasks.ConclusionsWe derived and evaluated traditional machine learning and deep learning classifiers to identify vaping-related relevant, commercial, and provape tweets. Overall, deep learning classifiers such as LSTM-CNN had superior performance and had the added advantage of requiring no preprocessing. The performance of these classifiers supports the development of a vaping surveillance system.

Project description:BACKGROUND:In plants, long non-protein coding RNAs are believed to have essential roles in development and stress responses. However, relative to advances on discerning biological roles for long non-protein coding RNAs in animal systems, this RNA class in plants is largely understudied. With comparatively few validated plant long non-coding RNAs, research on this potentially critical class of RNA is hindered by a lack of appropriate prediction tools and databases. Supervised learning models trained on data sets of mostly non-validated, non-coding transcripts have been previously used to identify this enigmatic RNA class with applications largely focused on animal systems. Our approach uses a training set comprised only of empirically validated long non-protein coding RNAs from plant, animal, and viral sources to predict and rank candidate long non-protein coding gene products for future functional validation. RESULTS:Individual stochastic gradient boosting and random forest classifiers trained on only empirically validated long non-protein coding RNAs were constructed. In order to use the strengths of multiple classifiers, we combined multiple models into a single stacking meta-learner. This ensemble approach benefits from the diversity of several learners to effectively identify putative plant long non-coding RNAs from transcript sequence features. When the predicted genes identified by the ensemble classifier were compared to those listed in GreeNC, an established plant long non-coding RNA database, overlap for predicted genes from Arabidopsis thaliana, Oryza sativa and Eutrema salsugineum ranged from 51 to 83% with the highest agreement in Eutrema salsugineum. Most of the highest ranking predictions from Arabidopsis thaliana were annotated as potential natural antisense genes, pseudogenes, transposable elements, or simply computationally predicted hypothetical protein. Due to the nature of this tool, the model can be updated as new long non-protein coding transcripts are identified and functionally verified. CONCLUSIONS:This ensemble classifier is an accurate tool that can be used to rank long non-protein coding RNA predictions for use in conjunction with gene expression studies. Selection of plant transcripts with a high potential for regulatory roles as long non-protein coding RNAs will advance research in the elucidation of long non-protein coding RNA function.

Dataset Information

A Genetic Attack Against Machine Learning Classifiers to Steal Biometric Actigraphy Profiles from Health Related Sensor Data.

Publications

A Genetic Attack Against Machine Learning Classifiers to Steal Biometric Actigraphy Profiles from Health Related Sensor Data.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets