- nexa - server-nexa.polito.it

Stop Explaining Black Box ML Models for High Stakes Decisions and UseInterpretable Models Instead (2019)
by Giacomo Tesio Aug. 12, 2020

Aug. 12, 2020

Black box machine learning models are currently being used for high stakes decision-making throughout society, causing problems throughout healthcare, criminal justice, and in other domains. People have hoped that creating methods for explaining these black box models will alleviate some of these problems, but trying to explain black box models, rather than creating models that are interpretablein the first place, is likely to perpetuate bad practices and can potentially cause catastrophic harm to society. There is a way forward – it isto design models that are inherently interpretable. This manuscript clarifies the chasm between explaining black boxes and using inherently interpretable models, outlines several key reasons why explainable blackboxes should be avoided in high-stakes decisions, identifies challenges to interpretable machine learning, and provides several example applications where interpretable models could potentially replace black box models in criminal justice, healthcare, and computer vision. Continua su https://arxiv.org/pdf/1811.10154.pdf Giacomo

1 0

A cosa serve la Blockchain? [was: No to Blockchain Credentials of COVID-19 Test Results for Entry to Public Spaces]
by Giacomo Tesio Aug. 12, 2020

Aug. 12, 2020

La blockchain trasforma l'ignoranza informatica diffusa in denaro vero attraverso lo spreco di grandi quantità di energia elettrica. La trasformazione avviene attraverso tre metodologie principali: - schemi Ponzi math-washed su larga scala (anche noti come criptovalute) - distrazione di finanziamenti (pubblici e/o privati) da attività di ricerca più serie ed utili alla società - contratti di consulenza (e simili) come esperti di blockchain (sviluppatori, ma anche avvocati, economisti etc...) L'unico uso efficace ed efficiente della blockchain ad oggi noto all'uomo consiste nella realizzazione di filtri automatici per scartare i curricula sulla base delle parole chiave associate (smart contract etc..) On Sun, 08 Aug 2020 14:32:01 +0000 Davide Carboni <dcarboni(a)gmail.com> wrote: > la blockchain (poi vediamo cos'è) non ha > nessun senso quando esiste un'autorità che può certificare un dato con > un banale PKI. In generale, la blockchain non ha alcun senso quando esiste una entità esterna che garantisce la funzione off-chain per cui è adottata. Un esempio? Come moneta. La blockchain non ha senso come moneta perché è sempre uno Stato a garantire (o NON garantire) la validità dei trasferimenti di proprietà (e la proprietà stessa ;-). Ed è sempre uno Stato a battere la moneta che viene usata come riferimento stabile per i prezzi. > La blockchain è uno scomodo artifact che risulta dall'esecuzione di un > protocollo di consenso basato su proof of work ovvero un consumo di corrente e banda crescente al crescere dell'utenza ovvero tanto meno sostenibile quanto più è diffusa ovvero l'esatto contrario della scalabilità ovvero un enorme incentivo a centralizzare di fatto il mining (eliminando la necessità di un consenso DISTRIBUITO! :-D) > ed incentivo economico > e che serve a risolvere un problema abbastanza ben formulato, ovvero > come ottenere il consenso eh già... il famoso consenso distribuito della blockchain. Quello che la rende immutabile! Tanto immutabile che a Luglio 2016 i miner di Ethereum per non perdere faccia e soldi... l'hanno modificata. Così. Come se non fosse immutabile. Come avranno fatto? Ah già... è immutabile per chi CREDE che lo sia! ;-) > sullo stato di un sistema distribuito in > cui i partecipanti sono non autenticati... e le "permissione blockchain"? (ricorda, lo schema Ponzi è solo una delle metodologie della blockchain) > dove nessuno ha un ruolo di master e Libra? (ricorda, lo schema Ponzi è solo una delle metodologie della blockchain) > e dove potenzialmente tutti i partecipanti sono malevoli potenzialmente sì, ma fior di statistiche nonché recenti esperimenti effettuati su uno dei maggiori social network mondiali evidenziano che per la stragrande maggioanza... ...sono fessi. ;-) > ma non coordinati fra loro. ..fin quando non aderiscono ad uno dei pochi mining pool esistenti. Ora, senza voler entrare in truffe più elaborate come Tether, tutte le criptovalute basate sulla blockchain sono allucinazioni collettive: i token rappresentano la promessa che sarà ancora possibile trovare un fesso a cui cederle in cambio di una contropartita (in denaro o altro). Il "valore" di quei byte dipende dal numero di allucinati e dalla fiducia cieca che questi hanno nella loro allucinazione. A breve sociologi e psicologi inizieranno a studiarle applicando categorie analoghe a quelle usate per il Tempio del Popolo a Jonestown. > Con questi requisiti estremi il Nakamoto consensus fornisce una > soluzione che sta su da 10 anni. E personalmente sono ancora convinto che non vedrà il 2021. Naturalmente potrei sbagliarmi di qualche mese, ma sono in tanti ad aver notato come i ratti stiano abbandonando la nave: da Telegram a Facebook, per non parlare delle vendite milionarie fatte per sottrarre il poco denaro rimasto nel sistema. > Ora che questo non sia un problema pratico > ma solo teorico è opinabile Veramente quello che descrivi non è nemmeno un problema teorico. "P = NP?" è un problema teorico. Quello che descrivi è marketing: un problema immaginario progettato per giustificare la proposta di una soluzione. > forse non è un problema per lei La blockchain? Per me non è un problema, ma il sintomo fastidioso di un problema grave. Il suo successo deriva dall'ignoranza informatica diffusa: un problema enorme per chiunque se ne occupi professionalmente. > ma per alcuni può esserlo. Diciamo piuttosto che tanti possono essere convinti di avere questo problema (o quanto meno di essere circondati di fessi che compreranno quei byte a più di quanto li hanno pagati loro). D'altronde anche l'omeopatia ha i propri adepti. > In fondo che bisogno c'era di inventare l'aereo quando potevamo > prendere il treno. Aereo e treno svolgono in modo diverso una funzione utile: viaggiare. Che funzione svolge la blockchain? (propriamente detta, con Merkle tree, PoW etc..) Qualche tempo fa ho proposto ad un esperto di blockchain di descrivermi che funzione svolgeva quest'ultima nel suo business affinché gli proponessi (gratuitamente) una soluzione alternativa più semplice ed efficace: https://twitter.com/giacomotesio/status/1282308279255937026 Come puoi leggere, non è stato difficile. Di nuovo l'altro giorno per Stefano ho abbozzato un protocollo molto semplice in grado di fornire le stesse garanzie della blockchain (e altre, con qualche piccola estensione). Ora, fare il BlockShame Buster è divertente, ma se deve diventare un mestiere a tempo pieno voglio essere pagato! :-D Tuttavia, qualunque sia l'uso non Ponziano (definirle "speculative" sarebbe nobilitare le criptovalute) della blockchain, esiste SEMPRE una soluzione tecnica migliore che fornisce tutte le garanzie necessarie senza sprecare corrente. Chiunque non è interessato ad una o più delle tre metodologie di truffa indicate all'inizio, farebbe bene a tenersi ben lontano da questa fanta-tecnologia. Giacomo

3 3

The Global Internet Is a Mirage - The New York Times
by Alberto Cammozzo Aug. 12, 2020

Aug. 12, 2020

<https://www.nytimes.com/2020/08/10/technology/global-internet.html> The U.S. government’s proposed ban on Chinese apps like TikTok and WeChat plays into technologists’ fears that the internet utopia is crumbling. The worry is that instead of a world brought closer together by the internet, a tech fight between the United States and China threatens to further splinter the digital world along country borders. I share those concerns. But let me explain why a splintered internet isn’t so novel, or necessarily a horrible thing. First, the internet was never as global or interconnected as the ideal. What we mean when we talk about a unified global internet is a history in which the internet was dominated by America, with U.S. companies and U.S. values infusing the world. The exception was China, which operated a parallel internet world. For years, foreign governments at times pushed back at the American-tinged internet. They sometimes had understandable reasons. Germany, for example, has strong norms of personal privacy and strict rules against denial of the Holocaust. That has resulted in conflict with the American internet companies’ standards of personal data collection and free expression. Other times, governments have imposed restrictions on online activity to silence opposition from their own citizens. Whether or not we agree with such tactics, the internet has never been a single global blob where borders didn’t matter. And do we want it to be? I’m an American, and I prefer our relatively freewheeling internet to what exists in Russia or Vietnam. But I also recognize that each country has its own tax codes, labor laws and auto safety regulations. When Ford makes car bumpers, it has to figure out how to alter designs to meet different safety rules in Italy and Nigeria. There are technical reasons that it’s trickier to make country-to-country rules about a website than the strength of car bumpers. But the idea of internet policy changing when you go from Brazil to Argentina is not crazy. That’s not to say that there’s nothing to worry about. I’m concerned that banning apps, writing laws restricting what people can say online or shutting down internet access entirely costs people digital lifelines to the outside world, and that the internet is one more way for authoritarian regimes to exert dominance. But it’s not productive to pine for a utopian internet that never really existed. When technologists lament the fracturing of the internet world, I wonder if what they’re really mourning is the fracturing of the world, period.

3 3

Facial recognition as the Manhattan Project of our time
by don Luca Peyron Aug. 12, 2020

Aug. 12, 2020

Rather than revising or expanding legislation, we could perhaps benefit more from having more inner dialogues. Along with an alarm that ideally would go off when we resort to fallacy and moral disengagement when engaging in critical self-analysis. Does an algorithm or app already exist for this? https://innovationorigins.com/facial-recognition-as-the-manhattan-project-o… _________________________ don Luca Peyron Direttore Ufficio per la Pastorale Universitaria Arcidiocesi di Torino www.universitari.to.it via XX settembre 83, Torino tel. 011 5156239

1 0

The EU is launching a market for personal data. Here’s what that means for privacy.
by Giacomo Tesio Aug. 12, 2020

Aug. 12, 2020

In a radical shift for the EU’s data governance strategy, the Trusts Project promotes data sharing as a civic duty. The new European data governance strategy (pdf) takes a fundamentally different approach. With it, the EU will become an active player in facilitating the use and monetization of its citizens’ personal data. Unveiled by the European Commission in February 2020, the strategy outlines policy measures and investments to be rolled out in the next five years. This new strategy represents a radical shift in the EU’s focus, from protecting individual privacy to promoting data sharing as a civic duty. Specifically, it will create a pan-European market for personal data through a mechanism called a data trust. A data trust is a steward that manages people’s data on their behalf and has fiduciary duties toward its clients. The EU’s new plan considers personal data to be a key asset for Europe. However, this approach raises some questions. First, the EU’s intent to profit from the personal data it collects puts European governments in a weak position to regulate the industry. Second, the improper use of data trusts can actually deprive citizens of their rights to their own data. The Trusts Project, the first initiative put forth by the new EU policies, will be implemented by 2022. With a €7 million budget, it will set up a pan-European pool of personal and nonpersonal information that should become a one-stop shop for businesses and governments looking to access citizens’ information. Global technology companies will not be allowed to store or move Europeans’ data. Instead, they will be required to access it via the trusts. Citizens will collect “data dividends,” which haven’t been clearly defined but could include monetary or nonmonetary payments from companies that use their personal data. With the EU’s roughly 500 million citizens poised to become data sources, the trusts will create the world’s largest data market. For citizens, this means the data created by them and about them will be held in public servers and managed by data trusts. The European Commission envisions the trusts as a way to help European businesses and governments reuse and extract value from the massive amounts of data produced across the region, and to help European citizens benefit from their information. The project documentation, however, does not specify how individuals will be compensated. Data trusts were first proposed by internet pioneer Sir Tim Berners Lee in 2018, and the concept has drawn considerable interest since then. Just like the trusts used to manage one’s property, data trusts may serve different purposes: they can be for-profit enterprises, or they can be set up for data storage and protection, or to work for a charitable cause. IBM and Mastercard have built a data trust to manage the financial information of their European clients in Ireland; the UK and Canada have employed data trusts to stimulate the growth of the AI industries there; and recently, India announced plans to establish its own public data trust to spur the growth of technology companies. Continua su https://www.technologyreview.com/2020/08/11/1006555/eu-data-trust-trusts-pr… Spero francamente che sia un pesce d'aprile. Giacomo

2 1

Changing World, Changing Mozilla
by Giacomo Tesio Aug. 11, 2020

Aug. 11, 2020

Over the last while, it has been clear that Mozilla is not structured properly to create these new things — and to build the better internet we all deserve. Today we announced a significant restructuring of Mozilla Corporation. This will strengthen our ability to build and invest in products and services that will give people alternatives to conventional Big Tech. Sadly, the changes also include a significant reduction in our workforce by approximately 250 people. [...] I desperately wish there was some other way to set Mozilla up for long term success in building a better internet. But to go further, we must be organized to be able to think about a different world. To imagine that technology will become embedded in our world even more than it is, and we want that technology to have different characteristics and values than we experience today. So going forward we will be smaller. We’ll also be organizing ourselves very differently, acting more quickly and nimbly. We’ll experiment more. We’ll adjust more quickly. We’ll join with allies outside of our organization more often and more effectively. We’ll meet people where they are. We’ll become great at expressing and building our core values into products and programs that speak to today’s issues. We’ll join and build with all those who seek openness, decency, empowerment and common good in online life. [...] I also realize this vision will feel abstract to many. With this in mind, we have mapped out five specific areas to focus on as we roll out this new structure over the coming months: 1. New focus on product... 2. New mindset... 3. New focus on technology... 4. New focus on community... 5. New focus on economics... We’re fortunate that Firefox and Mozilla retain a high degree of trust in the world. Trust and a feeling of authenticity feel unusual in tech today. But there is a sense that people want more from us. They want to work with us, to build with us. The changes we are making today are hard. But with these changes we believe we’ll be ready to meet these people — and the challenges and opportunities facing the future of the internet — head on. https://blog.mozilla.org/blog/2020/08/11/changing-world-changing-mozilla/ Giacomo

1 0

first ‘FLY AI’ report on the current use and future potential of artificial intelligence in aviation
by don Luca Peyron Aug. 11, 2020

Aug. 11, 2020

The first Fly AI report provides an overview of the many ways that artificial intelligence is already applied in our industry and assesses its potential to transform the sector. https://www.eurocontrol.int/publication/fly-ai-report _________________________ don Luca Peyron Direttore Ufficio per la Pastorale Universitaria Arcidiocesi di Torino www.universitari.to.it via XX settembre 83, Torino tel. 011 5156239

1 0

Il buon Samaritano e la sharing economy
by don Luca Peyron Aug. 10, 2020

Aug. 10, 2020

Semplice, diretto. Buona lettura dl p.s. anche se confonde il buon samaritano con S. Martino... https://www.avvenire.it/economia/pagine/il-buon-samaritano-e-la-sharing-eco… <https://www.avvenire.it/economia/pagine/il-buon-samaritano-e-la-sharing-eco…> _________________________ don Luca Peyron Direttore Ufficio per la Pastorale Universitaria Arcidiocesi di Torino www.universitari.to.it via XX settembre 83, Torino tel. 011 5156239

1 0

8 million people, 14 alerts: why some covid-19 apps are staying silent
by Giacomo Tesio Aug. 10, 2020

Aug. 10, 2020

When France launched its app for digital contact tracing, it looked like a possible breakthrough for the virus-ravaged country. After going live in June, StopCovid was downloaded by 2 million people in a short time, and digital affairs minister Cédric O said that “from the first downloads, the app helps avoid contamination, illness, and so deaths.” But officials soon had to walk their enthusiasm back after it emerged that in its first three weeks the app had alerted only 14 people to tell them they might have been exposed to the coronavirus. [...] Part of the criticism may be due to too much hype. [...] On a mathematical level, too, the low level of notifications might be expected [...] However, even with the most optimistic lens, it’s clear there’s a gulf between what was promised and what these apps are delivering. [...] “This is the prime risk for developers: you make one mistake and wipe out somebody’s battery,” says Andrew Eland, who until recently worked as an engineering director at Google and then DeepMind Health. [...] “If you want to know the best way to spend time and money on technology to track and trace coronavirus infections, it would probably be better to focus on making manual contact tracing more efficient,” says Eland. Continua su https://www.technologyreview.com/2020/07/10/1005027/8-million-people-14-ale… Giacomo

1 0

Lo sviluppo gentile
by don Luca Peyron Aug. 10, 2020

Aug. 10, 2020

Il cambio d’epoca che stiamo attraversando è prodotto dalla tecnologia e dal suo impatto sul nostro modo di comprendere noi e la realtà. Tuttavia il mondo della tecnologia è oggi descritto dalla categoria dell’innovazione. Se continueremo a guardare la tecnologia solamente come innovazione rischiamo però di non riuscirne a percepire la portata di trasformazione sociale né di orientarne verso il bene gli effetti. Buona lettura dl https://www.osservatoreromano.va/it/news/2020-08/lo-sviluppo-gentile.html _________________________ don Luca Peyron Direttore Ufficio per la Pastorale Universitaria Arcidiocesi di Torino www.universitari.to.it via XX settembre 83, Torino tel. 011 5156239

2 1