The Experts below are selected from a list of 88776 Experts worldwide ranked by ideXlab platform

Yana Welinder - One of the best experts on this subject based on the ideXlab platform.

  • a face tells more than a thousand posts developing face recognition privacy in social Networks
    Harvard Journal of Law & Technology, 2012
    Co-Authors: Yana Welinder
    Abstract:

    I. INTRODUCTION During the "Green Revolution" in 2009, the Iranian military posted photos from the protests on a website and invited citizens to identify twenty individual faces that were singled out in those photos. (1) They claimed to have arrested at least two of the individuals in the photos shortly after the protests. (2) According to some sources, the Iranian government tried to use face recognition technology to identify protesters, though its technology was still under development. (3) Imagine if the government could simply match these faces against the hundreds of billions of photos available on Facebook. The matches could reveal not only the protesters' names, (4) but also their whereabouts, their contacts, their online conversations with other protesters, and potentially their future plans. Faces are particularly good for identification purposes because they are distinctive and, in most cases, publicly visible. other personal features that are in plain sight--like coats and haircuts--can easily be replaced, but significantly altering a face to make it unrecognizable is difficult. And yet most people can remain anonymous, even in public, because they have only a limited set of acquaintances that can recognize them. The use of face recognition technology in social Networks shifts this paradigm. It can connect an otherwise anonymous face not only to a name--of which there can be several--but also to all the information in a social Network Profile. Given the risks of face recognition technology when combined with the vast amount of personal information aggregated in social Networks, this Article presents two central ideas. First, applying Professor Helen Nissenbaum's theory of contextual integrity, (5) I argue that face recognition technology in social Networks needs to be carefully regulated because it transforms the information that users share (e.g., it transforms a simple photo into biometric data that automatically identifies users) and provides this personally identifying information to new recipients beyond the user's control. Second, I identify the deficiencies in the current law and argue that law alone cannot solve this problem. A blanket prohibition on automatic face recognition in social Networks would stifle the development of these technologies, which are useful in their own right. At the same time, our traditional privacy framework of notice and consent cannot protect users who do not understand the automatic face recognition process and recklessly continue sharing their personal information due to strong Network effects. Instead, I propose a multifaceted solution aimed at lowering the costs of switching between social Networks and providing users with better information about how their data is used. (6) My argument is that once users are truly free to switch Networks, they will be able to exercise their choice to demand that social Networks respect their privacy expectations. In Part II, this Article begins with a general overview of face recognition technology and how it is implemented on Facebook. Part III of the Article uses the theory of contextual integrity to examine how social Networks may violate user privacy when they apply face recognition technology to user photos. It also explains why more traditional privacy theories--epitomized by Warren and Brandeis' right to be let alone--cannot address this problem because they are mostly concerned with the privacy of physical spaces and confidential information. Having identified how face recognition technology violates privacy in this context, Part III explains why a complete prohibition of face recognition technology or related data processing could prevent the development of useful technologies. This sets the stage for my multifaceted proposal. Part IV reviews current laws that could potentially apply to this problem and concludes that they do not offer sufficient privacy protection. Finally, Part V outlines a combination of legal, architectural, market, and norm-driven solutions that I believe could offer adequate privacy protection without unduly stifling innovation. …

  • a face tells more than a thousand posts developing face recognition privacy in social Networks
    Social Science Research Network, 2012
    Co-Authors: Yana Welinder
    Abstract:

    What is so special about a face? It is the one personally identifiable feature that we all show in public. Faces are particularly good for identification purposes because, unlike getting a new coat or haircut, significantly altering a face to make it unrecognizable is difficult. But since most people have only a limited set of acquaintances, they can often remain anonymous when doing something personal by themselves — even in public. The use of face recognition technology in social Networks shifts this paradigm. It can connect an otherwise anonymous face not only to a name — of which there can be several — but to all the information in a social Network Profile, including one’s friends, work and education history, status updates, and so forth.In this Article, I present two central ideas. First, applying the theory of contextual integrity, I argue that the current use face recognition technology in social Networks violates users’ privacy by changing the information that they share (from a simple photo to automatically identifying biometric data) and providing this information to new recipients beyond the users’ control. Second, I identify the deficiencies in the current law and argue that law alone cannot solve this problem. A blanket prohibition on automatic face recognition in social Networks would stifle the development of these technologies, which are useful in their own right. But at the same time, our traditional privacy framework of notice and consent cannot protect users who do not understand the automatic face recognition process and recklessly continue sharing their personal information due to strong Network effects. Instead, I propose a multifaceted solution aimed at lowering switching costs between social Networks and providing users with better information about how their data is used. My argument is that once users are truly free to leave, they will be able to exercise their choice in a meaningful way to demand that social Networks respect their privacy expectations.Though this Article specifically addresses the use of face recognition technology in social Networks, the proposed solution can be applied to other privacy problems arising in online platforms that accumulate personal information. More broadly, the undertaking to open up social Networks and make them more transparent and interoperable could address the concern that these Networks threaten to fragment the Web and lock in our personal information.

Richard Bonneau - One of the best experts on this subject based on the ideXlab platform.

  • netquilt deep multispecies Network based protein function prediction using homology informed Network similarity
    Bioinformatics, 2021
    Co-Authors: Meet Barot, Vladimir Gligorijevic, Kyunghyun Cho, Richard Bonneau
    Abstract:

    MOTIVATION Transferring knowledge between species is challenging: different species contain distinct proteomes and cellular architectures, which cause their proteins to carry out different functions via different interaction Networks. Many approaches to protein functional annotation use sequence similarity to transfer knowledge between species. These approaches cannot produce accurate predictions for proteins without homologues of known function, as many functions require cellular context for meaningful prediction. To supply this context, Network-based methods use protein-protein interaction (PPI) Networks as a source of information for inferring protein function and have demonstrated promising results in function prediction. However, most of these methods are tied to a Network for a single species, and many species lack biological Networks. RESULTS In this work, we integrate sequence and Network information across multiple species by computing IsoRank similarity scores to create a meta-Network Profile of the proteins of multiple species. We use this integrated multispecies meta-Network as input to train a maxout neural Network with Gene Ontology terms as target labels. Our multispecies approach takes advantage of more training examples, and consequently leads to significant improvements in function prediction performance compared to two Network-based methods, a deep learning sequence-based method, and the BLAST annotation method used in the Critial Assessment of Functional Annotation. We are able to demonstrate that our approach performs well even in cases where a species has no Network information available: when an organism's PPI Network is left out we can use our multi-species method to make predictions for the left-out organism with good performance. AVAILABILITY The code is freely available at https://github.com/nowittynamesleft/NetQuilt. SUPPLEMENTARY INFORMATION Supplementary data are available at Bioinformatics online.

  • netquilt deep multispecies Network based protein function prediction using homology informed Network similarity
    bioRxiv, 2020
    Co-Authors: Meet Barot, Vladimir Gligorijevic, Kyunghyun Cho, Richard Bonneau
    Abstract:

    Transferring knowledge between species is challenging: different species contain distinct proteomes and cellular architectures, which cause their proteins to carry out different functions via different interaction Networks. Many approaches to proteome and biological Network functional annotation use sequence similarity to transfer knowledge between species. These similarity-based approaches cannot produce accurate predictions for proteins without homologues of known function, as many functions require cellular or organismal context for meaningful function prediction. In order to supply this context, Network-based methods use protein-protein interaction (PPI) Networks as a source of information for inferring protein function and have demonstrated promising results in function prediction. However, the majority of these methods are tied to a Network for a single species, and many species lack biological Networks. In this work, we integrate sequence and Network information across multiple species by applying an IsoRank-derived Network alignment algorithm to create a meta-Network Profile of the proteins of multiple species. We then use this integrated multispecies meta-Network as input features to train a maxout neural Network with Gene Ontology terms as target labels. Our multispecies approach takes advantage of more training examples, and more diverse examples from multiple organisms, and consequently leads to significant improvements in function prediction performance. Further, we evaluate our approach in a setting in which an organism9s PPI Network is left out, using other organisms9 Network information and sequence homology in order to make predictions for the left-out organism, to simulate cases in which a newly sequenced species has no Network information available.

Mario Lemes Proenca - One of the best experts on this subject based on the ideXlab platform.

  • autonomous Profile based anomaly detection system using principal component analysis and flow analysis
    Applied Soft Computing, 2015
    Co-Authors: Gilberto Fernandes, Joel J P C Rodrigues, Mario Lemes Proenca
    Abstract:

    An original anomaly detection system using principal component analysis is proposed.Our system was evaluated using real traffic data from a university.PCA proved effective in creating a digital signature of Network traffic.Results pertaining to false alarm and accuracy rate are encouraging.Network anomalies were efficiently identified by our approach. Different techniques and methods have been widely used in the subject of automatic anomaly detection in computer Networks. Attacks, problems and internal failures when not detected early may badly harm an entire Network system. Thus, an autonomous anomaly detection system based on the statistical method principal component analysis (PCA) is proposed. This approach creates a Network Profile called Digital Signature of Network Segment using Flow Analysis (DSNSF) that denotes the predicted normal behavior of a Network traffic activity through historical data analysis. That digital signature is used as a threshold for volume anomaly detection to detect disparities in the normal traffic trend. The proposed system uses seven traffic flow attributes: bits, packets and number of flows to detect problems, and source and destination IP addresses and Ports, to provides the Network administrator necessary information to solve them. Via evaluation techniques performed in this paper using real Network traffic data, results showed good traffic prediction by the DSNSF and encouraging false alarm generation and detection accuracy on the detection schema using thresholds.

Meet Barot - One of the best experts on this subject based on the ideXlab platform.

  • netquilt deep multispecies Network based protein function prediction using homology informed Network similarity
    Bioinformatics, 2021
    Co-Authors: Meet Barot, Vladimir Gligorijevic, Kyunghyun Cho, Richard Bonneau
    Abstract:

    MOTIVATION Transferring knowledge between species is challenging: different species contain distinct proteomes and cellular architectures, which cause their proteins to carry out different functions via different interaction Networks. Many approaches to protein functional annotation use sequence similarity to transfer knowledge between species. These approaches cannot produce accurate predictions for proteins without homologues of known function, as many functions require cellular context for meaningful prediction. To supply this context, Network-based methods use protein-protein interaction (PPI) Networks as a source of information for inferring protein function and have demonstrated promising results in function prediction. However, most of these methods are tied to a Network for a single species, and many species lack biological Networks. RESULTS In this work, we integrate sequence and Network information across multiple species by computing IsoRank similarity scores to create a meta-Network Profile of the proteins of multiple species. We use this integrated multispecies meta-Network as input to train a maxout neural Network with Gene Ontology terms as target labels. Our multispecies approach takes advantage of more training examples, and consequently leads to significant improvements in function prediction performance compared to two Network-based methods, a deep learning sequence-based method, and the BLAST annotation method used in the Critial Assessment of Functional Annotation. We are able to demonstrate that our approach performs well even in cases where a species has no Network information available: when an organism's PPI Network is left out we can use our multi-species method to make predictions for the left-out organism with good performance. AVAILABILITY The code is freely available at https://github.com/nowittynamesleft/NetQuilt. SUPPLEMENTARY INFORMATION Supplementary data are available at Bioinformatics online.

  • netquilt deep multispecies Network based protein function prediction using homology informed Network similarity
    bioRxiv, 2020
    Co-Authors: Meet Barot, Vladimir Gligorijevic, Kyunghyun Cho, Richard Bonneau
    Abstract:

    Transferring knowledge between species is challenging: different species contain distinct proteomes and cellular architectures, which cause their proteins to carry out different functions via different interaction Networks. Many approaches to proteome and biological Network functional annotation use sequence similarity to transfer knowledge between species. These similarity-based approaches cannot produce accurate predictions for proteins without homologues of known function, as many functions require cellular or organismal context for meaningful function prediction. In order to supply this context, Network-based methods use protein-protein interaction (PPI) Networks as a source of information for inferring protein function and have demonstrated promising results in function prediction. However, the majority of these methods are tied to a Network for a single species, and many species lack biological Networks. In this work, we integrate sequence and Network information across multiple species by applying an IsoRank-derived Network alignment algorithm to create a meta-Network Profile of the proteins of multiple species. We then use this integrated multispecies meta-Network as input features to train a maxout neural Network with Gene Ontology terms as target labels. Our multispecies approach takes advantage of more training examples, and more diverse examples from multiple organisms, and consequently leads to significant improvements in function prediction performance. Further, we evaluate our approach in a setting in which an organism9s PPI Network is left out, using other organisms9 Network information and sequence homology in order to make predictions for the left-out organism, to simulate cases in which a newly sequenced species has no Network information available.

Gilberto Fernandes - One of the best experts on this subject based on the ideXlab platform.

  • autonomous Profile based anomaly detection system using principal component analysis and flow analysis
    Applied Soft Computing, 2015
    Co-Authors: Gilberto Fernandes, Joel J P C Rodrigues, Mario Lemes Proenca
    Abstract:

    An original anomaly detection system using principal component analysis is proposed.Our system was evaluated using real traffic data from a university.PCA proved effective in creating a digital signature of Network traffic.Results pertaining to false alarm and accuracy rate are encouraging.Network anomalies were efficiently identified by our approach. Different techniques and methods have been widely used in the subject of automatic anomaly detection in computer Networks. Attacks, problems and internal failures when not detected early may badly harm an entire Network system. Thus, an autonomous anomaly detection system based on the statistical method principal component analysis (PCA) is proposed. This approach creates a Network Profile called Digital Signature of Network Segment using Flow Analysis (DSNSF) that denotes the predicted normal behavior of a Network traffic activity through historical data analysis. That digital signature is used as a threshold for volume anomaly detection to detect disparities in the normal traffic trend. The proposed system uses seven traffic flow attributes: bits, packets and number of flows to detect problems, and source and destination IP addresses and Ports, to provides the Network administrator necessary information to solve them. Via evaluation techniques performed in this paper using real Network traffic data, results showed good traffic prediction by the DSNSF and encouraging false alarm generation and detection accuracy on the detection schema using thresholds.