Tipo de contenido

Next Web: web 3.0, web semántica y el futuro de internet > Procesamiento del lenguaje natural

    sortFiltrar Ordenar
    4 resultados

    Página Web


    Publicado el 29.11.2018 por Equipo GNOSS

    ¿Es más complicado que las máquinas entiendan el español que otros idiomas?

    En este artículo de Xataka se recogen opiniones de expertos en procesamiento de lenguaje natural, en las que exponen las dificultades de los sistemas para reconocer contextos y procesar ambigüedades, localismos o ironías.

    Según el artículo, esto es más complicado en español que en inglés y mucho más en lenguas con una morfología más compleja, como por ejemplo el euskera, el alemán o el finés con sus mecanismos de palabras compuestas.


    Página Web


    Publicado el 21.11.2018 por Equipo GNOSS

    Finalmente, una máquina que puede terminar su oración (New York Times)

    Completar el pensamiento de otra persona no es un una tarea sencilla para la Inteligencia Artificial. Pero los nuevos sistemas están empezando a descifrar el código del lenguaje natural.

    Este artículo de NYT detalla que Google reveló recientemene un sistema llamado Bert  que puede adivinar las palabras que faltan en millones de oraciones  (como "el hombre entró a una tienda y compró un ____ de leche"). También puede comprender muchas de las relaciones fundamentales entre las palabras en el idioma inglés, según afirma Jacob Devlin,  el investigador de Google al cargo de la creación de Bert.

    Bert tuvo éxito en parte porque se apoyaba en una enorme capacidad de procesamiento informático que no estaba disponible en las redes neuronales de años anteriores. Analizó todos los artículos de Wikipedia a lo largo de varios días utilizando docenas de procesadores construidos por Google específicamente para entrenar redes neuronales .

    Este mes, Google "abre la fuente" de su sistema Bert,  una vez entrenado en 102 idiomas, para que otros puedan aplicarlo a tareas adicionales. 




    Publicado el 1.10.2018 por Equipo GNOSS

    Artificial Intelligence and Life in 2030. Stanford University

    "Artificial Intelligence and Life in 2030" One Hundred Year Study on Artificial Intelligence: Report of the 2015-2016 Study Panel, Stanford University, Stanford, CA,  September 2016. Doc: http://ai100.stanford.edu/2016-report. Accessed:  September 6, 2016.

    Executive Summary. Artificial Intelligence (AI) is a science and a set of computational technologies that are inspired by—but typically operate quite differently from—the ways people use their nervous systems and bodies to sense, learn, reason, and take action. While the rate of progress in AI has been patchy and unpredictable, there have been significant advances since the field's inception sixty years ago. Once a mostly academic area of study, twenty-first century AI enables a constellation of mainstream technologies that are having a substantial impact on everyday lives. Computer vision and AI planning, for example, drive the video games that are now a bigger entertainment industry than Hollywood. Deep learning, a form of machine learning based on layered representations of variables referred to as neural networks, has made speech-understanding practical on our phones and in our kitchens, and its algorithms can be applied widely to an array of applications that rely on pattern recognition. Natural Language Processing (NLP) and knowledge representation and reasoning have enabled a machine to beat the Jeopardy champion and are bringing new power to Web searches.

    While impressive, these technologies are highly tailored to particular tasks. Each application typically requires years of specialized research and careful, unique construction. In similarly targeted applications, substantial increases in the future uses of AI technologies, including more self-driving cars, healthcare diagnostics and targeted treatments, and physical assistance for elder care can be expected. AI and robotics will also be applied across the globe in industries struggling to attract younger workers, such as agriculture, food processing, fulfillment centers, and factories. They will facilitate delivery of online purchases through flying drones, self-driving trucks, or robots that can get up the stairs to the front door.

    This report is the first in a series to be issued at regular intervals as a part of the One Hundred Year Study on Artificial Intelligence (AI100). Starting from a charge given by the AI100 Standing Committee to consider the likely influences of AI in a typical North American city by the year 2030, the 2015 Study Panel, comprising experts in AI and other relevant areas focused their attention on eight domains they considered most salient: transportation; service robots; healthcare; education; low-resource communities; public safety and security; employment and workplace; and entertainment. In each of these domains, the report both reflects on progress in the past fifteen years and anticipates developments in the coming fifteen years. Though drawing from a common source of research, each domain reflects different AI influences and challenges, such as the difficulty of creating safe and reliable hardware (transportation and service robots), the difficulty of smoothly interacting with human experts (healthcare and education), the challenge of gaining public trust (low-resource communities and public safety and security), the challenge of overcoming fears of marginalizing humans (employment and workplace), and the social and societal risk of diminishing interpersonal interactions (entertainment). The report begins with a reflection on what constitutes Artificial Intelligence, and concludes with recommendations concerning AI-related policy. These recommendations include accruing technical expertise about AI in government and devoting more resources—and removing impediments—to research on the fairness, security, privacy, and societal impacts of AI systems.

    Contrary to the more fantastic predictions for AI in the popular press, the Study Panel found no cause for concern that AI is an imminent threat to humankind. No machines with self-sustaining long-term goals and intent have been developed, nor are they likely to be developed in the near future. Instead, increasingly useful applications of AI, with potentially profound positive impacts on our society and economy are likely to emerge between now and 2030, the period this report considers. At the same time, many of these developments will spur disruptions in how human labor is augmented or replaced by AI, creating new challenges for the economy and society more broadly. Application design and policy decisions made in the near term are likely to have long-lasting influences on the nature and directions of such developments, making it important for AI researchers, developers, social scientists, and policymakers to balance the imperative to innovate with mechanisms to ensure that AI's economic and social benefits are broadly shared across society. If society approaches these technologies primarily with fear and suspicion, missteps that slow AI's development or drive it underground will result, impeding important work on ensuring the safety and reliability of AI technologies. On the other hand, if society approaches AI with a more open mind, the technologies emerging from the field could profoundly transform society for the better in the coming decades.

    Study Panel: 

    Peter Stone, Chair, University of Texas at Austin
    Rodney Brooks, Rethink Robotics
    Erik Brynjolfsson, Massachussets Institute of Technology
    Ryan Calo, University of Washington
    Oren Etzioni, Allen Institute for AI
    Greg Hager, Johns Hopkins University
    Julia Hirschberg, Columbia University
    Shivaram Kalyanakrishnan, Indian Institute of Technology Bombay
    Ece Kamar, Microsoft Research
    Sarit Kraus, Bar Ilan University
    Kevin Leyton-Brown, University of British Columbia
    David Parkes, Harvard University
    William Press, University of Texas at Austin
    AnnaLee (Anno) Saxenian, University of California, Berkeley
    Julie Shah, Massachussets Institute of Technology
    Milind Tambe, University of Southern California
    Astro Teller, X




    Página Web


    Publicado el 26.9.2018 por Equipo GNOSS

    WatsomApp: Inteligencia artificial para mejorar la convivencia en las aulas

    La empresa española KIO–AI ha desarrollado una herramienta tecnológica, basada en la inteligencia artificial Watson de IBM, para identificar y prevenir casos de acoso escolar. El programa, bautizado como WatsomApp, combina juegos on line y conversaciones con robots humanoides (QBO y Snow) para detectar conflictos en las aulas e identificar a alumnos que pueden estar siendo hostigados y a los posibles acosadores, así como a los líderes y a aquellos que están más aislados y pueden ser más vulnerables.