Warned That Big, Messy AI Systems Would Generate Racist, Unfair Results

Total Page:16

File Type:pdf, Size:1020Kb

Warned That Big, Messy AI Systems Would Generate Racist, Unfair Results JULY/AUG 2021 | DON’T BE EVIL warned that big, messy AI systems would generate racist, unfair results. Google brought her in to prevent that fate. Then it forced her out. Can Big Tech handle criticism from within? BY TOM SIMONITE NEW ROUTES TO NEW CUSTOMERS E-COMMERCE AT THE SPEED OF NOW Business is changing and the United States Postal Service is changing with it. We’re offering e-commerce solutions from fast, reliable shipping to returns right from any address in America. Find out more at usps.com/newroutes. Scheduled delivery date and time depend on origin, destination and Post Office™ acceptance time. Some restrictions apply. For additional information, visit the Postage Calculator at http://postcalc.usps.com. For details on availability, visit usps.com/pickup. The Okta Identity Cloud. Protecting people everywhere. Modern identity. For one patient or one billion. © 2021 Okta, Inc. and its affiliates. All rights reserved. ELECTRIC WORD WIRED 29.07 I OFTEN FELT LIKE A SORT OF FACELESS, NAMELESS, NOT-EVEN- A-PERSON. LIKE THE GPS UNIT OR SOME- THING. → 38 ART / WINSTON STRUYE 0 0 3 FEATURES WIRED 29.07 “THIS IS AN EXTINCTION EVENT” In 2011, Chinese spies stole cybersecurity’s crown jewels. The full story can finally be told. by Andy Greenberg FATAL FLAW How researchers discovered a teensy, decades-old screwup that helped Covid kill. by Megan Molteni SPIN DOCTOR Mo Pinel’s bowling balls harnessed the power of physics—and changed the sport forever. by Brendan I. Koerner HAIL, MALCOLM Inside Roblox, players built a fascist Roman Empire. Did that make them fashies IRL? by Cecilia D’Anastasio REVEL’S BUMPY RIDE When the shared electric mopeds hit the streets, people went wild. Then came the crashes. by Kate Knibbs THE MANHATTAN PROJECT One building will P.38 “GET THAT MONEY!” house Earth’s scariest animal pathogens. by Geoff Manaugh and Nicola Twilley Jeffrey Fang started driving for Lyft seven years ago and quickly mastered the ins and outs of the gig economy. The THE EXILE hustle became an addiction. Then his minivan got carjacked When Google cut ties during a DoorDash delivery—with his young children inside. with Timnit Gebru, by Lauren Smiley it called into question its whole system of self-scrutiny. by Tom Simonite 0 0 5 CONTENTS WIRED 29.07 ELECTRIC MIND WORD GRENADES P.8 Rants & Raves P.11 Glenn Gould’s Pandemic Moment by Virginia Heffernan ON THE P.16 All In on the Crypto Fight by Paul Ford COVER P.20 The Hidden Beauty of Chemistry by Meghan Herbst P.22 No Graphics Card, No Minecraft, No Connection by Jennifer Berney P.24 The Sound of Money by Arielle Pardes P.26 Translucent Nostalgia by Cecilia D’Anastasio P.28 Zoom Towns Rising by Aarian Marshall P.30 When AI Pulls the Trigger When photographer by Will Knight Djeneba Aduayom saw P.32 Cloud Support: Should I Use Timnit Gebru wearing an App to Slow Down? a blue scarf traditional by Meghan O’Gieblyn among some ethnic groups in Gebru’s native Ethiopia, P.34 The Best Ebikes she knew it would be in her by Adrienne So portrait. Aduayom draped the scarf around her. “It was just a perfect balance,” she says, “of mixing multi- SIX-WORD ple layers, physically and emotionally.” SCI-FI P.128 Very Short Stories by WIRED readers 0 0 6 What else can we do for our children? Help make the world more sustainable? And our portfolio too? For some of life’s questions, you’re not alone. Together we can find an answer. ubs.com/lifegoals Sustainable investing strategies aim to incorporate environmental, social and governance (ESG) considerations into investment process and portfolio construction. Strategies across geographies and styles approach ESG analysis and incorporate the findings in a variety of ways. The returns on portfolios consisting primarily of sustainable investments may be lower or higher than portfolios where ESG factors, exclusions, or other sustainability issues are not considered, and the investment opportunities available to such portfolios may also differ. As a firm providing wealth management services to clients, UBS Financial Services Inc. offers investment advisory services in its capacity as an SEC-registered investment adviser and brokerage services in its capacity as an SEC-registered broker-dealer. Investment advisory services and brokerage services are separate and distinct, differ in material ways and are governed by different laws and separate arrangements. It is important that clients understand the ways in which we conduct business, that they carefully read the agreements and disclosures that we provide to them about the products or services we offer. For more information, please review the PDF document at ubs.com/relationshipsummary. © UBS 2021. All rights reserved. UBS Financial Services Inc. is a subsidiary of UBS AG. Member FINRA/SIPC. ELECTRIC WORD WIRED 29.07 RANTS AND RAVES In June, Andy Greenberg dished up a scoop about one couple’s quest to fix the notoriously broken ice cream machines at McDonald’s. Gilad Edelman revisited the rationale behind Section 230, the landmark law that shields major tech corporations from liability over defamatory content on their platforms. And on WIRED.com (and page 62 in this issue), Megan Molteni writes about a scientist and a tenacious graduate student who unearthed an old scientific error with devastating consequences for the world’s response to Covid-19. correct the record. Perhaps west, if someone sneezes or RE: “SACRED COMMANDMENT/ that’s what WIRED really wants? coughs, it starts out as small ↙ FALSE IDOL” To go back to the old days, droplets but dries as it travels where if you wanted to com- from the body to aerosol size. Once an internet platform is ment on things you had to get Kudos to the researchers for aware that a posting is false and approval from editors at big their persistence, especially Readers defamatory, it should immedi- publishing houses? I prefer the the forensic investigator. complain ately delete it and note the reason open internet, and I thank Sec- —Mike, via [email protected] about for doing so. Neither the poster tion 230 for enabling it to exist nor the platform should escape the way it does. —@mmasnick, We often talk about the lack defamation, responsibility for the harm they via Twitter of dialog between science and appliances, have caused. —Robert Dawes, via public policy. Equally at fault is and siloed [email protected] RE: “SOFT-SERVE the lack of dialog between sci- HARDBALL” entific disciplines. That aero- science: We know it’s possible to have sol engineers and public health an internet without Section 230, I’ve often noticed that McDon- specialists and epidemiolo- insurrections, and a compromised gists were talking past each democracy, because other parts ald’s can’t keep its soft-serve machines working, but I didn’t other only underscores the of the world manage it. The US dangers of siloed academic version of the internet isn’t a pre- expect that sad fact to metas- tasize into a full-blown scan- thinking. We need people with requisite for accountability. skills to connect the dots. —@neilpbrady, via Twitter dal. —@harrymccracken, via Twitter —Junaid Ahmad, via mail@ WIRED.com As Section 230 advocate Cathy Gellis notes, in the “old days,” This story seems to be a pre-internet, you needed the reflection of the current appli- This is a great example of how approval of a gatekeeper—such ance manufacturing industry. important it is to teach histori- as a large magazine—to get your They make substandard prod- cal thinking and research skills. voice out there. The internet, and ucts, charge extra for service It’s a particular kind of ques- Section 230 with it, are part of agreements that ought not to tioning and literacy that, when RE: “FATAL FLAW” what makes it possible for her to be necessary, and use parts overlapped with other disci- that fail in two or three years. plines, is key to human sur- A complete abuse of what vival. —@homeonSaunders, used to be “engineering.” via Twitter —Werner Hofstatter, via mail@ WIRED.com Can’t believe an article on sci- “One of the better entific papers and research If McDonald’s truly had its could be this thrilling. franchisees’ interests at heart, —@CalibanCrew, via Twitter examples of just it would want the ice cream machines running as often as possible. Why don’t they? They how pernicious should have been negotiating with Taylor to get the Kytch on every machine. —Dan Grasso, confirmation bias via [email protected] RE: “FATAL FLAW” GET MORE WIRED can be.” All WIRED stories can be found As an infection control officer online, but only subscribers —@Roessler, via Twitter at a public safety agency, I get unlimited access. If you are have always been suspicious already a print subscriber, you of the droplet versus aerosol can authenticate your account argument. In the arid South- at WIRED.com/register. 0 0 8 NO PURCHASE NECESSARY. A PURCHASE WILL NOT INCREASE YOUR CHANCES OF WINNING. VOID IN MA, MI, VA AND WHERE PROHIBITED BY LAW. The Winston Off-Road Revival Sweepstakes starts on 6/1/21 and ends on 9/30/21. Open only to legal residents of the 50 U.S. and DC (excluding residents of MA, MI, and VA) who are smokers, 21 years of age or older. For Official Rules including entry instructions, how to participate without redeeming points, odds of winning and prize details visit https://winstonrewards.prizelogic.com/rules. Sponsor: ITG Brands, LLC, 714 Green Valley Road, Greensboro, NC 27408. Ford Motor Company (i) will have no liability or responsibility for any claim arising in connection with participation in this Sweepstakes or the award of the Prize and (ii) has not offered or sponsored this Sweepstakes in any way.
Recommended publications
  • Identificação De Textos Em Imagens CAPTCHA Utilizando Conceitos De
    Identificação de Textos em Imagens CAPTCHA utilizando conceitos de Aprendizado de Máquina e Redes Neurais Convolucionais Relatório submetido à Universidade Federal de Santa Catarina como requisito para a aprovação da disciplina: DAS 5511: Projeto de Fim de Curso Murilo Rodegheri Mendes dos Santos Florianópolis, Julho de 2018 Identificação de Textos em Imagens CAPTCHA utilizando conceitos de Aprendizado de Máquina e Redes Neurais Convolucionais Murilo Rodegheri Mendes dos Santos Esta monografia foi julgada no contexto da disciplina DAS 5511: Projeto de Fim de Curso e aprovada na sua forma final pelo Curso de Engenharia de Controle e Automação Prof. Marcelo Ricardo Stemmer Banca Examinadora: André Carvalho Bittencourt Orientador na Empresa Prof. Marcelo Ricardo Stemmer Orientador no Curso Prof. Ricardo José Rabelo Responsável pela disciplina Flávio Gabriel Oliveira Barbosa, Avaliador Guilherme Espindola Winck, Debatedor Ricardo Carvalho Frantz do Amaral, Debatedor Agradecimentos Agradeço à minha mãe Terezinha Rodegheri, ao meu pai Orlisses Mendes dos Santos e ao meu irmão Camilo Rodegheri Mendes dos Santos que sempre estiveram ao meu lado, tanto nos momentos de alegria quanto nos momentos de dificuldades, sempre me deram apoio, conselhos, suporte e nunca duvidaram da minha capacidade de alcançar meus objetivos. Agradeço aos meus colegas Guilherme Cornelli, Leonardo Quaini, Matheus Ambrosi, Matheus Zardo, Roger Perin e Victor Petrassi por me acompanharem em toda a graduação, seja nas disciplinas, nos projetos, nas noites de estudo, nas atividades extracurriculares, nas festas, entre outros desafios enfrentados para chegar até aqui. Agradeço aos meus amigos de infância Cássio Schmidt, Daniel Lock, Gabriel Streit, Gabriel Cervo, Guilherme Trevisan, Lucas Nyland por proporcionarem momentos de alegria mesmo a distância na maior parte da caminhada da graduação.
    [Show full text]
  • Vikas Sindhwani Google May 17-19, 2016 2016 Summer School on Signal Processing and Machine Learning for Big Data
    Real-time Learning and Inference on Emerging Mobile Systems Vikas Sindhwani Google May 17-19, 2016 2016 Summer School on Signal Processing and Machine Learning for Big Data Abstract: We are motivated by the challenge of enabling real-time "always-on" machine learning applications on emerging mobile platforms such as next-generation smartphones, wearable computers and consumer robotics systems. On-device models in such settings need to be highly compact, and need to support fast, low-power inference on specialized hardware. I will consider the problem of building small-footprint non- linear models based on kernel methods and deep learning techniques, for on-device deployments. Towards this end, I will give an overview of various techniques, and introduce new notions of parsimony rooted in the theory of structured matrices. Such structured matrices can be used to recycle Gaussian random vectors in order to build randomized feature maps in sub-linear time for approximating various kernel functions. In the deep learning context, low-displacement structured parameter matrices admit fast function and gradient evaluation. I will discuss how such compact nonlinear transforms span a rich range of parameter sharing configurations whose statistical modeling capacity can be explicitly tuned along a continuum from structured to unstructured. I will present empirical results on mobile speech recognition problems, and image classification tasks. I will also briefly present some basics of TensorFlow: a open-source library for numerical computations on data flow graphs. Tensorflow enables large-scale distributed training of complex machine learning models, and their rapid deployment on mobile devices. Bio: Vikas Sindhwani is Research Scientist in the Google Brain team in New York City.
    [Show full text]
  • BRKIOT-2394.Pdf
    Unlocking the Mystery of Machine Learning and Big Data Analytics Robert Barton Jerome Henry Distinguished Architect Principal Engineer @MrRobbarto Office of the CTAO CCIE #6660 @wirelessccie CCDE #2013::6 CCIE #24750 CWNE #45 BRKIOT-2394 Cisco Webex Teams Questions? Use Cisco Webex Teams to chat with the speaker after the session How 1 Find this session in the Cisco Events Mobile App 2 Click “Join the Discussion” 3 Install Webex Teams or go directly to the team space 4 Enter messages/questions in the team space BRKIOT-2394 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 3 Tuesday, Jan. 28th Monday, Jan. 27th Wednesday, Jan. 29th BRKIOT-2600 BRKIOT-2213 16:45 Enabling OT-IT collaboration by 17:00 From Zero to IOx Hero transforming traditional industrial TECIOT-2400 networks to modern IoT Architectures IoT Fundamentals 08:45 BRKIOT-1618 Bootcamp 14:45 Industrial IoT Network Management PSOIOT-1156 16:00 using Cisco Industrial Network Director Securing Industrial – A Deep Dive. Networks: Introduction to Cisco Cyber Vision PSOIOT-2155 Enhancing the Commuter 13:30 BRKIOT-1775 Experience - Service Wireless technologies and 14:30 BRKIOT-2698 BRKIOT-1520 Provider WiFi at the Use Cases in Industrial IOT Industrial IoT Routing – Connectivity 12:15 Cisco Remote & Mobile Asset speed of Trains and Beyond Solutions PSOIOT-2197 Cisco Innovates Autonomous 14:00 TECIOT-2000 Vehicles & Roadways w/ IoT BRKIOT-2497 BRKIOT-2900 Understanding Cisco's 14:30 IoT Solutions for Smart Cities and 11:00 Automating the Network of Internet Of Things (IOT) BRKIOT-2108 Communities Industrial Automation Solutions Connected Factory Architecture Theory and 11:00 Practice PSOIOT-2100 BRKIOT-1291 Unlock New Market 16:15 Opening Keynote 09:00 08:30 Opportunities with LoRaWAN for IOT Enterprises Embedded Cisco services Technologies IOT IOT IOT Track #CLEMEA www.ciscolive.com/emea/learn/technology-tracks.html Cisco Live Thursday, Jan.
    [Show full text]
  • Gmail Smart Compose: Real-Time Assisted Writing
    Gmail Smart Compose: Real-Time Assisted Writing Mia Xu Chen∗ Benjamin N Lee∗ Gagan Bansal∗ [email protected] [email protected] [email protected] Google Google Google Yuan Cao Shuyuan Zhang Justin Lu [email protected] [email protected] [email protected] Google Google Google Jackie Tsay Yinan Wang Andrew M. Dai [email protected] [email protected] [email protected] Google Google Google Zhifeng Chen Timothy Sohn Yonghui Wu [email protected] [email protected] [email protected] Google Google Google Figure 1: Smart Compose Screenshot. ABSTRACT our proposed system design and deployment approach. This system In this paper, we present Smart Compose, a novel system for gener- is currently being served in Gmail. ating interactive, real-time suggestions in Gmail that assists users in writing mails by reducing repetitive typing. In the design and KEYWORDS deployment of such a large-scale and complicated system, we faced Smart Compose, language model, assisted writing, large-scale serv- several challenges including model selection, performance eval- ing uation, serving and other practical issues. At the core of Smart ACM Reference Format: arXiv:1906.00080v1 [cs.CL] 17 May 2019 Compose is a large-scale neural language model. We leveraged Mia Xu Chen, Benjamin N Lee, Gagan Bansal, Yuan Cao, Shuyuan Zhang, state-of-the-art machine learning techniques for language model Justin Lu, Jackie Tsay, Yinan Wang, Andrew M. Dai, Zhifeng Chen, Timothy training which enabled high-quality suggestion prediction, and Sohn, and Yonghui Wu. 2019. Gmail Smart Compose: Real-Time Assisted constructed novel serving infrastructure for high-throughput and Writing. In The 25th ACM SIGKDD Conference on Knowledge Discovery and real-time inference.
    [Show full text]
  • The Machine Learning Journey with Google
    The Machine Learning Journey with Google Google Cloud Professional Services The information, scoping, and pricing data in this presentation is for evaluation/discussion purposes only and is non-binding. For reference purposes, Google's standard terms and conditions for professional services are located at: https://enterprise.google.com/terms/professional-services.html. 1 What is machine learning? 2 Why all the attention now? Topics How Google can support you inyour 3 journey to ML 4 Where to from here? © 2019 Google LLC. All rights reserved. What is machine0 learning? 1 Machine learning is... a branch of artificial intelligence a way to solve problems without explicitly codifying the solution a way to build systems that improve themselves over time © 2019 Google LLC. All rights reserved. Key trends in artificial intelligence and machine learning #1 #2 #3 #4 Democratization AI and ML will be core Specialized hardware Automation of ML of AI and ML competencies of for deep learning (e.g., MIT’s Data enterprises (CPUs → GPUs → TPUs) Science Machine & Google’s AutoML) #5 #6 #7 Commoditization of Cloud as the platform ML set to transform deep learning for AI and ML banking and (e.g., TensorFlow) financial services © 2019 Google LLC. All rights reserved. Use of machine learning is rapidly accelerating Used across products © 2019 Google LLC. All rights reserved. Google Translate © 2019 Google LLC. All rights reserved. Why all the attention0 now? 2 Machine learning allows us to solve problems without codifying the solution. © 2019 Google LLC. All rights reserved. San Francisco New York © 2019 Google LLC. All rights reserved.
    [Show full text]
  • Google's 'Project Nightingale' Gathers Personal Health Data
    Google's 'Project Nightingale' Gathers Personal Health Data on Millions of Americans; Search giant is amassing health records from Ascension facilities in 21 states; patients not yet informed Copeland, Rob . Wall Street Journal (Online) ; New York, N.Y. [New York, N.Y]11 Nov 2019. ProQuest document link FULL TEXT Google is engaged with one of the U.S.'s largest health-care systems on a project to collect and crunch the detailed personal-health information of millions of people across 21 states. The initiative, code-named "Project Nightingale," appears to be the biggest effort yet by a Silicon Valley giant to gain a toehold in the health-care industry through the handling of patients' medical data. Amazon.com Inc., Apple Inc. and Microsoft Corp. are also aggressively pushing into health care, though they haven't yet struck deals of this scope. Share Your Thoughts Do you trust Google with your personal health data? Why or why not? Join the conversation below. Google began Project Nightingale in secret last year with St. Louis-based Ascension, a Catholic chain of 2,600 hospitals, doctors' offices and other facilities, with the data sharing accelerating since summer, according to internal documents. The data involved in the initiative encompasses lab results, doctor diagnoses and hospitalization records, among other categories, and amounts to a complete health history, including patient names and dates of birth. Neither patients nor doctors have been notified. At least 150 Google employees already have access to much of the data on tens of millions of patients, according to a person familiar with the matter and the documents.
    [Show full text]
  • Recreation Fun, Concerts in Full Swing in Sturbridge
    SPENCER FAMILY DENTAL Gentle Caring State of the Art Dentistry For The Whole Family Cosmetic Dentistry • Restorative Dentistry • Preventative Dentistry CROWNS • CAPS • BRIDGES • COMPLETE and PARTIAL DENTURES New We Strive NON SURGICAL GUM TREATMENT • ROOT CANAL THERAPY Patients SURGICAL SERVICES For Painless Welcome BREATH CLINIC-WE TREAT CHRONIC BAD BREATH Dentistry HERBAL DENTAL PRODUCTS • All Instruments Fully Sterilized • Most Insurances Accepted Dr. Nasser S. Hanna Conveniently Located On Route 9 • (Corner of Greenville St. & Main St.) 284 Main St., Spencer 508-885-5511 Mailed free to requesting homes in Sturbridge, Brimfield, Holland and Wales Vol. VII, No. 27 PROUD MEDIA SPONSOR OF RELAY FOR LIFE OF THE GREATER SOUTHBRIDGE AREA! COMPLIMENTARY HOME DELIVERY ONLINE: WWW.STURBRIDGEVILLAGER.NET Friday, July 11, 2014 THIS WEEK’S QUOTE Recreation fun, concerts “Be careful that victories do not in full swing in Sturbridge carry the seed of ARIOUS ACTIVITIES ON TURBRIDGE future defeats.” V S SUMMER SCHEDULE Ralph W. Sockman BY AMANDA COLLINS ers up the road, kids playing VILLAGER STAFF WRTIER catch while there are parents STURBRIDGE — If you drive sprawled out on a blanket as the past the Sturbridge town com- sun sets. mon on a Thursday night this Roll down the windows and INSIDE summer, you might see people take it slow to enjoy the sound of lugging lawn chairs and cool- live music from some the area’s ALMANAC ............. 2 most popular bands. Or better The Mad Bavarian Brass Band came yet, pull over and join in on the POLICE LOGS ......... 5 to play at the Sturbridge Town Sturbridge summertime tradi- OBITUARIES ..........
    [Show full text]
  • 321444 1 En Bookbackmatter 533..564
    Index 1 Abdominal aortic aneurysm, 123 10,000 Year Clock, 126 Abraham, 55, 92, 122 127.0.0.1, 100 Abrahamic religion, 53, 71, 73 Abundance, 483 2 Academy award, 80, 94 2001: A Space Odyssey, 154, 493 Academy of Philadelphia, 30 2004 Vital Progress Summit, 482 Accelerated Math, 385 2008 U.S. Presidential Election, 257 Access point, 306 2011 Egyptian revolution, 35 ACE. See artificial conversational entity 2011 State of the Union Address, 4 Acquired immune deficiency syndrome, 135, 2012 Black Hat security conference, 27 156 2012 U.S. Presidential Election, 257 Acxiom, 244 2014 Lok Sabha election, 256 Adam, 57, 121, 122 2016 Google I/O, 13, 155 Adams, Douglas, 95, 169 2016 State of the Union, 28 Adam Smith Institute, 493 2045 Initiative, 167 ADD. See Attention-Deficit Disorder 24 (TV Series), 66 Ad extension, 230 2M Companies, 118 Ad group, 219 Adiabatic quantum optimization, 170 3 Adichie, Chimamanda Ngozi, 21 3D bioprinting, 152 Adobe, 30 3M Cloud Library, 327 Adonis, 84 Adultery, 85, 89 4 Advanced Research Projects Agency Network, 401K, 57 38 42, 169 Advice to a Young Tradesman, 128 42-line Bible, 169 Adwaita, 131 AdWords campaign, 214 6 Affordable Care Act, 140 68th Street School, 358 Afghan Peace Volunteers, 22 Africa, 20 9 AGI. See Artificial General Intelligence 9/11 terrorist attacks, 69 Aging, 153 Aging disease, 118 A Aging process, 131 Aalborg University, 89 Agora (film), 65 Aaron Diamond AIDS Research Center, 135 Agriculture, 402 AbbVie, 118 Ahmad, Wasil, 66 ABC 20/20, 79 AI. See artificial intelligence © Springer Science+Business Media New York 2016 533 N.
    [Show full text]
  • Deep Learning on Mobile Devices – a Review
    Deep Learning on Mobile Devices – A Review Yunbin Deng FAST Labs, BAE Systems, Inc. Burlington MA 01803 ABSTRACT Recent breakthroughs in deep learning and artificial intelligence technologies have enabled numerous mobile applications. While traditional computation paradigms rely on mobile sensing and cloud computing, deep learning implemented on mobile devices provides several advantages. These advantages include low communication bandwidth, small cloud computing resource cost, quick response time, and improved data privacy. Research and development of deep learning on mobile and embedded devices has recently attracted much attention. This paper provides a timely review of this fast-paced field to give the researcher, engineer, practitioner, and graduate student a quick grasp on the recent advancements of deep learning on mobile devices. In this paper, we discuss hardware architectures for mobile deep learning, including Field Programmable Gate Arrays (FPGA), Application Specific Integrated Circuit (ASIC), and recent mobile Graphic Processing Units (GPUs). We present Size, Weight, Area and Power (SWAP) considerations and their relation to algorithm optimizations, such as quantization, pruning, compression, and approximations that simplify computation while retaining performance accuracy. We cover existing systems and give a state-of-the-industry review of TensorFlow, MXNet, Mobile AI Compute Engine (MACE), and Paddle-mobile deep learning platform. We discuss resources for mobile deep learning practitioners, including tools, libraries, models, and performance benchmarks. We present applications of various mobile sensing modalities to industries, ranging from robotics, healthcare and multi- media, biometrics to autonomous drive and defense. We address the key deep learning challenges to overcome, including low quality data, and small training/adaptation data sets.
    [Show full text]
  • Understanding Alphabet and Google, 2017
    This research note is restricted to the personal use of [email protected]. Understanding Alphabet and Google, 2017 Published: 24 February 2017 ID: G00297707 Analyst(s): Tom Austin, David Mitchell Smith, Yefim V. Natis, Isabelle Durand, Ray Valdes, Bettina Tratz-Ryan, Roberta Cozza, Daniel O'Connell, Lydia Leong, Jeffrey Mann, Andrew Frank, Brian Blau, Chris Silva, Mark Hung, Adam Woodyer, Matthew W. Cain, Steve Riley, Martin Reynolds, Whit Andrews, Alexander Linden, David Yockelson, Joe Mariano Google's size, market differentiation, rapid pace of innovation and ambitions can complicate fully understanding the vendor and its fit to current digital business needs. CIOs and IT leaders can use this report to explore in detail selected topics from the Gartner Vendor Rating. Key Findings ■ Two outcomes are apparent more than a year after the creation of the Alphabet-Google structure: Google is beginning to show increased momentum and has made significant investments in its enterprise offerings (most of its 2016 acquisitions were focused on this); and it is applying more discipline in Alphabet's "Other Bets." ■ Google is flourishing despite challenging external market factors: adverse publicity, competitors, government regulators and law enforcement. ■ Google values data, encourages bold investments in long-term horizons, pivots plans based on results in near real time, and reveres user-oriented engineering excellence. ■ Google is fully committed to 100% cloud-based and web-scale infrastructure, massive scaling, the maximum rate of change, and stream-lined business processes for itself and its customers. Recommendations CIOs and IT leaders managing vendor risk and performance should: ■ Plan for a long-term strategic relationship with Google based on an assumption that "what you see is what you get." Major vendor changes to core culture and fundamental operating principles in response to customer requests usually come slowly, if at all.
    [Show full text]
  • Repowered Feminist Analysis of Parks and Recreation
    Repowered Feminist Analysis of Parks and Recreation A Thesis submitted to Southern Utah University in partial fulfillment of the requirements for the degree of Master of Arts in Professional Communication December 2018 By Bailee M. Bahr Southern Utah University Thesis Committee: Kevin Stein, PhD, Chair I certify that I have read and viewed this project and that, in my opinion, it is satisfactory in scope and quality as a thesis for the degree of Master of Professional Communication. Repowered Feminist Analysis 1 Kevin Stein __________________________________ Kevin Stein, Ph.D., Capstone Chair Matthew Barton __________________________________ Matthew Barton, Ph.D., MAPC Graduate Director Abstract Repowered Feminist Analysis 2 This paper analyzes the television show Parks and Recreation in order to find principles of Foss and Foss’ (2009) characteristics of repowered feminism. This paper aims to discover if Leslie Knope represents a new form of feminism, what characteristics specifically that she represents, and if these qualities contribute to a freer, less oppressed Leslie Knope. The analysis examines three episodes of the show and uses feminist rhetorical criticism to analyze the findings. I found that repowered feminism applies both to a feminist’s concerns with feminist issues and the applicability of repowered feminism to all types of problem solving. Knope, whether focusing directly on feminist issues or on the various obstacles she faces while doing her job, is usually presented as more successful when she implements the characteristics of repowered feminism. Keywords: parks and recreation, repowered feminism, pop culture Acknowledgements A massive hug and kiss to my adorable husband who encouraged me to finish my thesis in spite of the plethora of excuses.
    [Show full text]
  • Exposing Safety Culture Neglect in Public Transit INTERNATIONAL OFFICERS LAWRENCE J
    OFFICIAL JOURNAL OF THE AMALGAMATED TRANSIT UNION|AFL-CIO/CLC SEPTEMBER / OCTOBER 2017 Exposing safety culture neglect in public transit INTERNATIONAL OFFICERS LAWRENCE J. HANLEY International President JAVIER M. PEREZ, JR. International Executive Vice President NEWSBRIEFS OSCAR OWENS International Secretary-Treasurer INTERNATIONAL VICE PRESIDENTS RICHARD M. MURPHY Newburyport, MA – [email protected] JANIS M. BORCHARDT Madison, WI – [email protected] PAUL BOWEN Canton, MI – [email protected] KENNETH R. KIRK Lancaster, TX – [email protected] MARCELLUS BARNES Flossmore, IL – [email protected] RAY RIVERA Lilburn, GA – [email protected] YVETTE TRUJILLO Thornton, CO – [email protected] GARY JOHNSON, SR. Cleveland, OH – [email protected] ROBIN WEST JIC Locals gather for collaborative training Halifax, NS – [email protected] More than 30 Locals from across the U.S. and Canada gathered at the Tommy Douglas Conference JOHN COSTA Center for an innovative Joint Industry Council training. The attendees heard from experts and staff Kenilworth, NJ – [email protected] on the new realities of bargaining and campaigning against large, deep-pocketed multinational CHUCK WATSON employers. The training included a comprehensive understanding of the RFP process, a breakdown Syracuse, NY – [email protected] of the revenue agreement, negotiations, strategies for building strength within our Locals, and CLAUDIA HUDSON Oakland, CA – [email protected] planning for organizing campaigns. The Locals also exchanged ideas and experiences in dealing with these companies that will change our strategic
    [Show full text]