November 2016 - nexa - server-nexa.polito.it

One by One, ISIS Social Media Experts Are Killed as Result of F.B.I. Program

by Alberto Cammozzo

<http://www.nytimes.com/2016/11/24/world/middleeast/isis-recruiters-social-m…> WASHINGTON — In the summer of 2015, armed American drones over eastern Syria stalked Junaid Hussain, an influential hacker and recruiter for the Islamic State. For weeks, Mr. Hussain was careful to keep his young stepson by his side, and the drones held their fire. But late one night, Mr. Hussain left an internet cafe alone, and minutes later a Hellfire missile killed him as he walked between two buildings in Raqqa, Syria, the Islamic State’s de facto capital. [] Initially the threat posed by the Legion was primarily seen as a problem for law enforcement officials. But as the threat worsened last year, and the F.B.I. stepped up the monitoring of terrorism suspects around the country, the bureau pressed the military to focus on the group, according to current and former American officials. While American and British forces conducted a series of drone strikes on members of the group, the F.B.I. sifted through thousands of the Legion’s followers on social media to figure out who had actually been inspired to take action. In the last two years, it has arrested nearly 100 people in cases involving the terrorist group. Several of the arrests were of people who had direct contact with the Legion. Many of the others involved were “folks who first came on our radar because we became aware of them” through their connections with Hussain and Reyaad Khan, also a British citizen, who was another leader of the group, according to Andrew McCabe, deputy director of the F.B.I. []

Nov. 28, 2016

Grave challenges remain before the promise of individually tailored medicine becomes reality.

by Antonio Vetrò

Breve articolo di Nature che riesce a toccare i diversi punti da risolvere prima di accettare entusiasticamente l’idea che basti una grosse mole di dati possa bastare ad aprire la promessa nuova era della medicina. http://www.nature.com/news/the-power-of-big-data-must-be-harnessed-for-medi…

Nov. 28, 2016

Children’s access to the internet: presenting recent findings | Global Kids Online

by J.C. DE MARTIN

http://blogs.lse.ac.uk/gko/childrens-access-to-the-internet-presenting-rece… (Sent from my wireless device; please excuse brevity and typos (if any))

Nov. 27, 2016

Fwd: [Creis_Terminal] appel à communicationsFwd: Colloque « Big data et visibilité en ligne, un enjeu pluridisciplinaire de l’économie numérique »

by Mariella Berra

-------- Messaggio Inoltrato -------- Oggetto: [Creis_Terminal] appel à communicationsFwd: Colloque « Big data et visibilité en ligne, un enjeu pluridisciplinaire de l’économie numérique » Data: Sun, 27 Nov 2016 12:34:06 +0100 Mittente: Genevieve Vidal <genevieve.vidal(a)univ-paris13.fr> A: CREIS <creis_terminal(a)lecreis.org> > chers collègues, > > Nous organisons en novembre 2017 un colloque international intitulé « > Big data et visibilité en ligne, un enjeu pluridisciplinaire de > l’économie numérique ». www.bigdata-lecolloque.org > Il est porté par 5 universités, 6 laboratoires et la MSH de Toulouse. > Cet évènement scientifique donnera lieu à la publication des actes du > colloque ainsi qu’à la publication d’un ouvrage collectif pour les > contributions qui seront retenues par le comité scientifique. > Je vous laisse le soin de découvrir l’appel à communication. Les > propositions de communication sont attendues pour le 30 janvier 2017. > Merci d’avance de faire tourner ce message dans vos réseaux > respectifs. > Au plaisir de vous retrouver prochainement. > > > Christophe Alcantara > IDETCOM/ UT1 Capitole > www.bigdata-lecolloque.org > > PS: Par avance, je vous prie de bien vouloir m'excuser pour les > possibles envois multiples

Nov. 27, 2016

Re: [nexa] Automated Inference on Criminality using Face Images

by Alberto Cammozzo

Certo, ormai il ML viene applicato indiscriminatamente in qualsiasi ambito con quasi nessuna competenza delle discipline e soprattutto con nessuna attenzione alle conseguenze. Nessuno sente il bisogno di capire *perché* le foto dei detenuti/pregiudicati e quelle degli incensurati liberi risultino diverse. Ma adottando il modello offerto dal ML si corre il rischio di rafforzare le cause (che restano ignote) per cui (forse) funziona. Certo il solo volto non porterà all'incriminazione di nessuno, ma le implicite previsioni fatte seguendo il modello contribuiranno ulteriormente alla costruzione del criminale prima che questo abbia mai commesso un crimine. E' la famosa profezia che si autoavvera di Robert K. Merton. Certo, si può dire che questi comportamenti discriminanti non sono una novità; ma non venivano chiamati scienza, ma pregiudizio. Il ML non richiede un modello che sia *esplicativo* di nessi causali, tuttavia produce un modello *descrittivo* (ma opaco), usabile come modello *predittivo* e una possibile pericolosa conseguenza è che finisca per diventare *prescrittivo*. Prima di adottare modelli ML in contesti socialmente rilevanti che posano avere conseguenze per i cittadini sarebbe ragionevole pretendere che i nessi causali dietro al modello vengano esplicitati e validati. Altrimenti qualsiasi scelta che si appoggi sul modello dovrebbe essere deprecata. Alberto On 11/26/2016 02:18 PM, Alessandro Mantelero wrote: > appunto e, come in altri casi, chi studia le nuove tecnologie dovrebbe > prima studiare un po' le passate esperienza/dottrine nel settore di > interesse. > > AM > > On Sat, 26 Nov 2016 12:12:12 +0100 > Alberto Cammozzo <ac+nexa(a)zeromx.net> wrote: >> La riscossa di Lombroso passa per la fisiognomica machine learned. >> >> <https://arxiv.org/abs/1611.04135> >> >> Automated Inference on Criminality using Face Images >> Xiaolin Wu, Xi Zhang >> (Submitted on 13 Nov 2016 (v1), last revised 21 Nov 2016 (this >> version, v2)) >> >> We study, for the first time, automated inference on criminality based >> solely on still face images. Via supervised machine learning, we build >> four classifiers (logistic regression, KNN, SVM, CNN) using facial >> images of 1856 real persons controlled for race, gender, age and facial >> expressions, nearly half of whom were convicted criminals, for >> discriminating between criminals and non-criminals. All four classifiers >> perform consistently well and produce evidence for the validity of >> automated face-induced inference on criminality, despite the historical >> controversy surrounding the topic. Also, we find some discriminating >> structural features for predicting criminality, such as lip curvature, >> eye inner corner distance, and the so-called nose-mouth angle. Above >> all, the most important discovery of this research is that criminal and >> non-criminal face images populate two quite distinctive manifolds. The >> variation among criminal faces is significantly greater than that of the >> non-criminal faces. The two manifolds consisting of criminal and >> non-criminal faces appear to be concentric, with the non-criminal >> manifold lying in the kernel with a smaller span, exhibiting a law of >> normality for faces of non-criminals. In other words, the faces of >> general law-biding public have a greater degree of resemblance compared >> with the faces of criminals, or criminals have a higher degree of >> dissimilarity in facial appearance than normal people. >> _______________________________________________ >> nexa mailing list >> nexa(a)server-nexa.polito.it >> https://server-nexa.polito.it/cgi-bin/mailman/listinfo/nexa > > -- > Prof. Avv. Alessandro Mantelero > Politecnico di Torino > > Nexa Center for Internet and Society | Director of Privacy > Politecnico di Torino–Tongji University| Coordinator, Double Degree > program in Management and IP Law > Nanjing University of Information Science and Technology | Part-time > Expert, School of Public Administration > European Data Protection Law Review | Associate Editor > > http://staff.polito.it/alessandro.mantelero > @mantelero > > EMAIL POLICY: once a day (Mon-Fri)

Nov. 26, 2016

Re: [nexa] Automated Inference on Criminality using Face Images

by Alessandro Mantelero

appunto e, come in altri casi, chi studia le nuove tecnologie dovrebbe prima studiare un po' le passate esperienza/dottrine nel settore di interesse. AM On Sat, 26 Nov 2016 12:12:12 +0100 Alberto Cammozzo <ac+nexa(a)zeromx.net> wrote: > La riscossa di Lombroso passa per la fisiognomica >machine learned. > > <https://arxiv.org/abs/1611.04135> > > Automated Inference on Criminality using Face Images > Xiaolin Wu, Xi Zhang > (Submitted on 13 Nov 2016 (v1), last revised 21 Nov 2016 >(this version, v2)) > > We study, for the first time, automated inference on >criminality based > solely on still face images. Via supervised machine >learning, we build > four classifiers (logistic regression, KNN, SVM, CNN) >using facial > images of 1856 real persons controlled for race, gender, >age and facial > expressions, nearly half of whom were convicted >criminals, for > discriminating between criminals and non-criminals. All >four classifiers > perform consistently well and produce evidence for the >validity of > automated face-induced inference on criminality, despite >the historical > controversy surrounding the topic. Also, we find some >discriminating > structural features for predicting criminality, such as >lip curvature, > eye inner corner distance, and the so-called nose-mouth >angle. Above > all, the most important discovery of this research is >that criminal and > non-criminal face images populate two quite distinctive >manifolds. The > variation among criminal faces is significantly greater >than that of the > non-criminal faces. The two manifolds consisting of >criminal and > non-criminal faces appear to be concentric, with the >non-criminal > manifold lying in the kernel with a smaller span, >exhibiting a law of > normality for faces of non-criminals. In other words, >the faces of > general law-biding public have a greater degree of >resemblance compared > with the faces of criminals, or criminals have a higher >degree of > dissimilarity in facial appearance than normal people. > _______________________________________________ > nexa mailing list > nexa(a)server-nexa.polito.it > https://server-nexa.polito.it/cgi-bin/mailman/listinfo/nexa -- Prof. Avv. Alessandro Mantelero Politecnico di Torino Nexa Center for Internet and Society | Director of Privacy Politecnico di Torino–Tongji University| Coordinator, Double Degree program in Management and IP Law Nanjing University of Information Science and Technology | Part-time Expert, School of Public Administration European Data Protection Law Review | Associate Editor http://staff.polito.it/alessandro.mantelero @mantelero EMAIL POLICY: once a day (Mon-Fri)

Nov. 26, 2016

Automated Inference on Criminality using Face Images

by Alberto Cammozzo

La riscossa di Lombroso passa per la fisiognomica machine learned. <https://arxiv.org/abs/1611.04135> Automated Inference on Criminality using Face Images Xiaolin Wu, Xi Zhang (Submitted on 13 Nov 2016 (v1), last revised 21 Nov 2016 (this version, v2)) We study, for the first time, automated inference on criminality based solely on still face images. Via supervised machine learning, we build four classifiers (logistic regression, KNN, SVM, CNN) using facial images of 1856 real persons controlled for race, gender, age and facial expressions, nearly half of whom were convicted criminals, for discriminating between criminals and non-criminals. All four classifiers perform consistently well and produce evidence for the validity of automated face-induced inference on criminality, despite the historical controversy surrounding the topic. Also, we find some discriminating structural features for predicting criminality, such as lip curvature, eye inner corner distance, and the so-called nose-mouth angle. Above all, the most important discovery of this research is that criminal and non-criminal face images populate two quite distinctive manifolds. The variation among criminal faces is significantly greater than that of the non-criminal faces. The two manifolds consisting of criminal and non-criminal faces appear to be concentric, with the non-criminal manifold lying in the kernel with a smaller span, exhibiting a law of normality for faces of non-criminals. In other words, the faces of general law-biding public have a greater degree of resemblance compared with the faces of criminals, or criminals have a higher degree of dissimilarity in facial appearance than normal people.

Nov. 26, 2016

Giornalismo: l'aggiornamento di un articolo on line deve essere subito visibile

by Alessandro Mantelero

L'aggiornamento di un articolo pubblicato on line deve essere immediatamente visibile al lettore, sia nel titolo sia nel contenuto dell'anteprima (la cosiddetta preview). Non è sufficiente apporre una postilla alla fine dell'articolo. Testo della decisione: http://www.garanteprivacy.it/web/guest/home/docweb/-/docweb-display/docweb/… Nota breve di commento Nel provvedimento il Garante sposa l'interpretazione del d.lgs. 196/2003 data a suo tempo dalla Cassazione(Cass. n. 5525 del 5 aprile 2012), che infatti il Garante espressamente richiama. Mi spiace, ma è un'interpretazione che personalmente non condivido e che crea discutibili oneri per le testate. Il punto centrale di questa argomentazione consiste nell'opinabile equiparazione fatta a suo tempo dalla Cassazione fra trattamento dati per finalità di conservazione di un archivio di giornale e trattamento dati nell'ambito dell'attività giornalistica di divulgazione di una notizia. Laddove infatti il giornalista narri eventi del passato, certamente la finalità informativa e ricostruttiva con cui le informazioni vengono trattate comporterà un’analisi che tenga anche conto dei fatti successivi ed integri una corretta rappresentazione dell’identità personale. E' invece diversa la fattispecie in cui non vi sia rievocazione di eventi, né esercizio dell’attività giornalistica o storica, ma semplicemente la raccolta delle edizioni di un giornale. In questo caso non si ha alcuna nuova pubblicazione, bensì solo conservazione documentale ed archivistica. In questa ipotesi la finalità è storico-archivistica, come tale attiene alla conservazione del passato, senza alterazione e come venuto via via in essere, e non alla sua rielaborazione o ricostruzione. Da qui la natura antitetica di un’eventuale opera di contestualizzazione che, secondo quanto affermato dalla Corte (e ora ripreso dal Garante), si traduca nell’integrazione e nella correlazione fra i documenti, attività che non a caso rientrano nel mestiere dello storico e vengono poste in essere nell’ambito della diversa finalità tesa alla ricostruzione degli eventi. Pretendere l’integrazione ad opera del titolare dell’archivio comporterebbe dunque un improprio «travaso» di competenze. D’altra parte proprio la stessa centralità della finalità del trattamento, invocata dalla Corte, conduce a tale conclusione: se lo scopo è quello di dar vita ad una raccolta dei materiali editi, la realizzazione di tale fine archivistico si traduce nella mera conservazione della notizia in quanto tale e non nella ricostruzione ex post degli eventi nella loro completezza; quest’ultima attiene infatti alla diversa finalità storica-ricostruttiva. Queste erano alcune considerazioni espresse a suo tempo sul tema e che mi paio ancora attuali, qui il testo integrale del commento: Mantelero, A. 2012. Right to Be Forgotten ed Archivi Storici Dei Giornali: La Cassazione Travisa il Diritto all'Oblio. La Nuova Giurisprudenza Civile Commentata 28(10) https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2176835 -- Prof. Avv. Alessandro Mantelero Politecnico di Torino Nexa Center for Internet and Society | Director of Privacy Politecnico di Torino–Tongji University| Coordinator, Double Degree program in Management and IP Law Nanjing University of Information Science and Technology | Part-time Expert, School of Public Administration European Data Protection Law Review | Associate Editor http://staff.polito.it/alessandro.mantelero @mantelero EMAIL POLICY: once a day (Mon-Fri)

Nov. 25, 2016

David Kaye su fake news e propaganda online

by fabio chiusi

Finalmente parole di saggezza in un dibattito che oscilla tra il dannoso, l'inutile e il pericoloso: https://freedex.org/2016/11/25/the-false-promise-of-banning-fake-news/ Quando Kaye parla del rischio di affidare ai governi, non ai social, la distinzione tra notizia vera e falsa, ricordo solo che proprio oggi Angela Merkel ha parlato di "regolamentare" i social media proprio su questo. (http://www.dw.com/en/german-social-bots-expert-to-brief-mps-on-fake-news-th…) Attenti a criminalizzare la propaganda. Del resto, come mi dice Philip Howard di Oxford sull'Espresso, ormai i bot politici li usano tutti. Il rischio censura è altissimo: http://espresso.repubblica.it/visioni/cultura/2016/11/25/news/cosi-i-govern… Buona serataf

Nov. 25, 2016

Re: [nexa] WashPost: "Russian propaganda effort helped spread ‘fake news’ during election, experts say"

by J.C. DE MARTIN

Dalla lista IP di David Farber. juan carlos * From: *Michael Robertson <mr(a)michaelrobertson.com <mailto:mr@michaelrobertson.com>> *Subject: **Re: [IP] Russian propaganda effort helped spread ‘fake news’ during election, experts say - The Washington Post* *Date: *November 25, 2016 at 8:55:36 AM EST *To: *David Farber <dave(a)farber.net <mailto:dave@farber.net>> *Cc: *ip <ip(a)listbox.com <mailto:ip@listbox.com>> The two examples they cite of fake news were actually not fake. 1) Hillary's health Due to citizen reporting we know that Hillary did have health issues. After public scrutiny, her campaign revised her condition first blaming allergies and layer pneumonia. 2) Paid anti Trump protesters We know from project veritas' undercover videos that operatives for the DNC paid mentally unstable people to disrupt Trump rallies and also people in Donald duck costumes. https://youtu.be/5IuJGHuIkzY https://youtu.be/EEQvsK5w-jY Parties in the videos either resigned or were fired after the videos were made public. It's entirely possible the Russians placed fake stories online to try and influence the election. However this article does not stand up to scrutiny. For the two instances references there's publicly available evidence that it's not fake. On 25/11/16 07:00, J.C. DE MARTIN wrote: > *Russian propaganda effort helped spread ‘fake news’ during election, > experts say* > > By Craig Timberg > > November 24 at 8:27 PM > > The flood of “fake news” this election season got support from a > sophisticated Russian propaganda campaign that created and spread > misleading articles online with the goal of punishing Democrat Hillary > Clinton, helping Republican Donald Trump and undermining faith in > American democracy, say independent researchers who tracked the operation. > > Russia’s increasingly sophisticated propaganda machinery — including > thousands of botnets, teams of paid human “trolls,” and networks of > websites and social-media accounts — echoed and amplified right-wing > sites across the Internet as they portrayed Clinton as a criminal > hiding potentially fatal health problems and preparing to hand control > of the nation to a shadowy cabal of global financiers. The effort also > sought to heighten the appearance of international tensions and > promote fear of looming hostilities with nuclear-armed Russia. > > Two teams of independent researchers found that the Russians exploited > American-made technology platforms to attack U.S. democracy at a > particularly vulnerable moment, as an insurgent candidate harnessed a > wide range of grievances to claim the White House. The sophistication > of the Russian tactics may complicate efforts by Facebook and Google > to crack down on “fake news,” as they have vowed to do after > widespread complaints about the problem. > > […] > > Continua qui: > https://www.washingtonpost.com/business/economy/russian-propaganda-effort-h… > > > > _______________________________________________ > nexa mailing list > nexa(a)server-nexa.polito.it > https://server-nexa.polito.it/cgi-bin/mailman/listinfo/nexa

Nov. 25, 2016