Download - Fronteiras da Inteligência Artificial
Fronteiras da Inteligência Artificial(ou IA sem fronteiras?)
Priscila Machado Vieira Lima, Ph.D.PESC/COPPE & NCE
Uma Homenagem a Antonia Jones"I never saw a wild thing sorry for itself.
A small bird will drop frozen dead from a boughWithout ever having felt sorry for itself."
David Herbert Lawrence (1885-1930)http://users.cs.cf.ac.uk/O.F.Rana/Antonia.J.Jones/
Professor Emerita of Neural and Evolutionary Computing
• Uma mudança de rumo na infância• Pólio ao 10 anos
• Educação em casa por um tempo
• O olhar para os outros • “Antonia is most fondly remembered by her friends and family as a patient, kind and thoughtful person
whose home was always open to stray animals and to humans in need of a meal, a listening ear and non-judgmental advice. This made her home chaotic during those times when "a wheel comes off," as she put it, but she always demonstrated patience and kindness toward others and maintained her dignity to the veryend of life.”
https://www.legacy.com/obituaries/staugustine/obituary.aspx?n=antonia-j-jones&pid=150288587
24/2/21 Programa de Engenharia de Sistemas e Computação 4
O ser humano Antonia Jane Jones (1943-2010)
• Formação• Dupla graduação em Matemática e Física, ambas com distinção, pela Reading University • Doutorado em Teoria dos Números na University of Cambridge
• Instituições• University of Nottingham; Institute for Advanced Study Princeton, New Jersey; Colorado
University, Boulder, Colorado; Royal Holloway College, University of London; Illinois State University, Normal, Illinois; Brunel University, West London; Imperial College, University of London, New Mexico State University, Las Cruces e School of Computer Science, Cardiff University, País de Gales.
24/2/21 Programa de Engenharia de Sistemas e Computação 6
Brevíssimo CV
• Livros técnicos• Mastering the Commodore 64, (com G. J. Carpenter). 384pp. Ellis Horwood Ltd. (1984)
• Consultorias a empresas• Consultoria sobre processamento de dados para o filme Rocinante (1986)
24/2/21 Programa de Engenharia de Sistemas e Computação 8
Antonia e a Tecnologia
• “Antonia was *the* most intelligent, creative and enthusiastic person I have come across. She will be missed by all her students and colleagues -- her sense of humour coupled with her desire to conquer human limits through technology and science will remain an inspiration ...”
Professor Omer Rana, 9/5/2011
24/2/21 Programa de Engenharia de Sistemas e Computação 9
A cientista Antonia
College Dean of International for the Physical Sciences and Engineering CollegeProfessor of Performance EngineeringSchool of Computer Science and InformaticsCardiff University(ex-orientado por Prof. Antonia Jones, 1998)
• “The Gamma Test [Evans 2002] is a non-linear modelling analysis tool that allows us to quantify the extent to which a numerical input/output data set can be expressed as a smooth relationship. In essence, it allows us to efficiently calculate that part of the variance of the output that cannot be accounted for by the existence of any smooth model based on the inputs, even though this model is unknown. A key aspect of this tool is its speed: the Gamma Test has time complexity O(Mlog M), where M is the number of data-points. For data sets consisting of a few thousand points and a reasonable number of attributes, a single run of the Gamma Test typically takes a few seconds. A simple introduction to the basic ideas is given in [Jones 2002]. The original commercial software winGammaTM is now Freeware and available from http://users.cs.cf.ac.uk/O.F.Rana/Antonia.J.Jones/GammaArchive/IndexPage.htm .”
• [Evans, 2002] D. Evans and Antonia J. Jones. A proof of the Gamma test. Proc. Roy. Soc. Lond. A 458:2759-2799. Preprint PDF [643 Kb]
• [Jones 2002] Antonia J. Jones, Dafydd Evans, Steve Margetts, Peter M. Durrant. The Gamma Test. Chapter IX in Heuristic and Optimization for Knowledge Discovery. Edited by by Ruhul Sarker, Hussein Abbass andCharles Newton. Idea Group Publishing, Hershey, PA. 2002. [464 Kb]
24/2/21 Programa de Engenharia de Sistemas e Computação 10
Gamma Test
• “The Gamma test computes the second moment of the noise distribution. Developments in 2002 (see Evans and Jones 2008 below) include new algorithms to compute as many higher moments of the noise distribution as is justified by the amount of available data. This enables an approximate reconstruction of the noise distribution - this may have interesting applications in communication theory and economic modelling for example.”
• [Evans 2008] Dafydd Evans and Antonia J. Jones Non-parametric estimation of residual moments andcovariance. Proc. Roy. Soc. Lond. Series A, 464(2099): 2831-2846, 2008. doi:10.1098/rspa.2007.0195. [Preprint PDF]
24/2/21 Programa de Engenharia de Sistemas e Computação 11
Gamma Test
• Atuação em Computação• Contribuições para análise de sistemas complexos e metaheurísticas
• “massively parallel models of computation, connectionist architectures; algorithms for the extraction of noise applied to the prediction and control of chaotic dynamics; analysis of probabilistic algorithms for hard combinatoric search; learning algorithms with particular reference to speech recognition and signal processing”
Resultaram de conhecimentos trazidos de outras áreas para a Computação
24/2/21 Programa de Engenharia de Sistemas e Computação 12
A cientista Antonia
Interdisciplinar desde a origem do termo...• Dartmouth Summer Research Project on Artificial
Intelligence (1956)• Marvin Minsky , Julian Bigelow, D.M. Mackay,. Ray
Solomonof, John Holland, John McCarthy, Claude Shannon, Nathanial Rochester, Oliver Selfridge, Allen Newell, Herbert Simon• Processamento de Linguagem Natural, Teoria da
Computação, Arquitetura de Computadores, Abstração, Criatividade…
http://jmc.stanford.edu/
13https://commons.wikimedia.org/w/index.php?title=Special:Search&search=john+mccarthy&fulltext=1&ns0=1&ns6=1&ns12=1&ns14=1&ns100=1&ns106=1#/media/File:John_McCarthy_Stanford.jpg
24/2/21 Programa de Engenharia de Sistemas e Computação
John McCarthy
Interdisciplinar desde a origem do termo...
Resolução de Problemas(Newell & Simon, McCarthy)
Teoria da Informação(Shannon, MacKay, Solomonoff)
Aprendizado de Máquina, Reconhecimanto de Padrões(Solomonoff, Rochester, Selfridge, Holland)
Cibernética, SW Básico e HW(Bigelow, Rochester)
Redes Neurais(Minsky, MacKay, Selfridge)
1424/2/21 Programa de Engenharia de Sistemas e Computação
IA na prática…
Representação
Adaptação
Busca
Programa de Engenharia de Sistemas e Computação 1524/2/21
IA na prática…
Lógica(s),Redes
SemânticasRedes Neurais
Aprendizadode
Máquina
Raciocínio
Resoluçãode
Problemas
Programa de Engenharia de Sistemas e Computação 1624/2/21
• Se eu já sei como fazer (programar)• Então
NÃO É IA!!!!!!!
24/2/21 Programa de Engenharia de Sistemas e Computação 17
A maldição...
https://www.pngegg.com/pt/png-ibdwo/download
Superando a maldição
• Aprendizado de Máquina• Modelos novos/aprimorados
• Aprendizado profundo• Redes convolucionais• Redes neurais sem pesos• MDPs• Árvores de Decisão• Programação lógica indutiva …
• Capacidade Computacional• Macro e micro
• Computadores multi-thread• Clusters• GPGPUs• FPGAs• Supercomputadores• AIoT• Computação quântica...
24/2/21 Programa de Engenharia de Sistemas e Computação 18
24/2/21 Programa de Engenharia de Sistemas e Computação 21
E novamente a interdisciplinaridade21
Cost-e!ective, Energy-e!icient, and Scalable StorageComputing for Large-scale AI Applications
JAEYOUNG DO, Microsoft Research, USAVICTOR C. FERREIRA, Federal University of Rio de Janeiro, BrazilHOSSEIN BOBARSHAD and MAHDI TORABZADEHKASHI, NGD Systems, USASIAVASH REZAEI and ALI HEYDARIGORJI, University of California, Irvine, USADIEGO SOUZA, Wespa Intelligent SystemsBRUNNO F. GOLDSTEIN and LEANDRO SANTIAGO, Federal University of Rio de JaneiroMIN SOO KIM, University of California, Irvine, USAPRISCILA M. V. LIMA and FELIPE M. G. FRANÇA, Federal University of Rio de Janeiro, BrazilVLADIMIR ALVES, NGD Systems, USA
The growing volume of data produced continuously in the Cloud and at the Edge poses signi!cant chal-lenges for large-scale AI applications to extract and learn useful information from the data in a timely ande"cient way. The goal of this article is to explore the use of computational storage to address such chal-lenges by distributed near-data processing. We describe Newport, a high-performance and energy-e"cientcomputational storage developed for realizing the full potential of in-storage processing. To the best of ourknowledge, Newport is the !rst commodity SSD that can be con!gured to run a server-like operating sys-tem, greatly minimizing the e#ort for creating and maintaining applications running inside the storage. Weanalyze the bene!ts of using Newport by running complex AI applications such as image similarity searchand object tracking on a large visual dataset. The results demonstrate that data-intensive AI workloads canbe e"ciently parallelized and o$oaded, even to a small set of Newport drives with signi!cant performancegains and energy savings. In addition, we introduce a comprehensive taxonomy of existing computationalstorage solutions together with a realistic cost analysis for high-volume production, giving a good big pictureof the economic feasibility of the computational storage technology.
CCS Concepts: • Information systems → Storage architectures; • Computer systems organization →Distributed architectures; • Computing methodologies → Arti!cial intelligence;
Additional Key Words and Phrases: Computational storage, in-storage processing, solid-state drive, similaritysearch, neural network, object tracking
This work is supported in part by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - Brasil (CAPES) -Finance Code 001. Also, this material is based upon work supported by the National Science Foundation under Grant No.1660071.Authors’ addresses: J. Do (Corresponding author), Microsoft Research, Redmond, Washington; email: [email protected]; V. C. Ferreira, B. F. Goldstein, L. Santiago, P. M. V. Lima, and F. M. G. França, Federal University of Riode Janeiro, Rio de Janeiro, Brazil; H. Bobarshad and M. Torabzadehkashi, NGD Systems, Irvine, California; S. Rezaei, A.HeydariGorji, and M. S. Kim, University of California, Irvine, Irvine, California; D. Souza, Wespa Intelligent Systems, Riode Janeiro, Brazil; V. Alves, NGD Systems, Irvine, California.Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without feeprovided that copies are not made or distributed for pro!t or commercial advantage and that copies bear this notice andthe full citation on the !rst page. Copyrights for components of this work owned by others than the author(s) must behonored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists,requires prior speci!c permission and/or a fee. Request permissions from [email protected].© 2020 Copyright held by the owner/author(s). Publication rights licensed to ACM.1553-3077/2020/10-ART21 $15.00https://doi.org/10.1145/3415580
ACM Transactions on Storage, Vol. 16, No. 4, Article 21. Publication date: October 2020.
Nem tudo são flores• Grandes conquistas recentes de IA
• 2005 – 2nd DARPA Grand Challenge – carros autônomos off-road• 2015 – AlphaGo derrota o campeão europeu de Go• 2020 – Skype Translator – 60-11 idiomas (voz) e entre 60 idiomas (texto)
• Mas erros acontecem…• 2017 – “Why Google’s Artificial Intelligence Confused a Turtle for a Rifle”• 2018 – “British parrot goes on shopping sprees through Alexa”• 2016 – “Tesla driver dies in first fatal crash while using autopilot mode” • 2018 – ” Self-driving Uber kills Arizona woman in first fatal crash involving pedestrian”• 2019 – “YouTube CEO defends site's recommendation system amid scrutiny”
• Responsável: Aprendizado de Máquina
A importância da Explicaçãp - Priscila M V Lima - CDIA - RioInfo - 2020 22
1. Desmistificação de IA2. IA centrada nos humanos3. Arquiteturas de computadores para IA4. IA na indústria5. IA na sociedade, na arte e na cultura6. IA Genérica
24/2/21 Programa de Engenharia de Sistemas e Computação 26
E as fronteiras?
Será preciso atravessá[email protected]
24/2/21 Programa de Engenharia de Sistemas e Computação 28
Será preciso atravessá-las!(pode dar uma espiadinha antes)
[email protected]@nce.ufrj.br
24/2/21 Programa de Engenharia de Sistemas e Computação 29