Thesis PROPUESTA DE GUÍA PRÁCTICA PARA LA APLICACIÓN DE TECNOLOGÍAS MACHINE LEARNING EN LAS EMPRESAS
Loading...
Date
2019-11-11
Authors
Journal Title
Journal ISSN
Volume Title
Program
MAGÍSTER EN INNOVACIÓN TECNOLÓGICA Y EMPRENDIMIENTO
Campus
Campus Vitacura, Santiago
Abstract
El machine learning es sin duda una tecnología que ha ido impactando de
forma exponencial en los aplicativos cotidianos y empresariales. Es nuestro deber
como líderes tecnológicos saber orientar bien los desafíos empresariales y darle
soluciones inteligentes. En este presente documento daremos un vistazo a que es
el machine learning y cómo podemos aplicar soluciones a los nuevos desafíos
empresariales en un mundo completamente real. Para interiorizarse y dar el primer
paso, no se requiere de grandes recursos ni expertos en el tema.
La tendencia en los últimos años, indica que algunas compañías apuestan por el
Machine Learning (ML) pero lo hacen por el momento y en gran medida en un
entorno controlado o de pruebas y sin influir en sus procesos de negocio principales
según variados estudios, se concluye que dar el salto a la aplicación en áreas de
negocio fundamentales es el gran reto al que deben hacer frente las organizaciones
en 2019.
Si bien esta tesina será una guía práctica y que permitirá contextualizar el grado de
complejidad para inicializarse, siempre debemos tener presente y como objetivo, las
siguientes preguntas:
¿Cómo evolucionar del Machine Learning en pequeña escala a una de mayor
alcance?
Esta respuesta se ha ido construyendo y generando una especie de tendencia, la
cual podemos resumir en estos 6 siguientes contextos:
1. Disponibilidad de datos bien gobernados. A medida que se avanza en el
Machine Learning más se entiende la necesidad de un gobierno del dato.
Data Governance ganará relevancia durante 2019 y, gracias a ello, los
profesionales podrán ser capaces de conocer bien los orígenes de datos e
identificar aquellos que son de calidad y los que no. Las máquinas de
aprendizaje se basan en datos de entrenamiento buenos, únicos y confiables,
por lo que es imprescindible aplicar una buena política de gobierno del dato.
2. Capacitación de los profesionales de las organizaciones para entender de
verdad las posibilidades del Machine Learning. Esto no sólo implica contratar
perfiles especializados, sino que conlleva también capacitar a los
trabajadores que conocen el negocio para que entiendan las posibilidades y
las limitaciones que ofrecen las técnicas de Machine Learning.
3. Desarrollo de nuevos perfiles especialistas en datos. Complementariamente
al punto anterior, la ampliación del alcance de Machine Learning, obligará a
que las compañías integren en sus equipos nuevos perfiles altamente
especializados, no solo el de Científico de Datos, sino, además:
a) Interlocutor de negocio con conocimientos de Machine Learning. Se
trata de un profesional que viene de las áreas de gestión de una empresa y toma decisiones estratégicas o tácitas y que, además,
cuenta con una formación en analítica suficiente para servir de enlace
entre los científicos de datos y las áreas de negocio.
b) Informático que sepa integrar algoritmos y modelos de ML. Un
trabajador que entienda cómo funcionan los algoritmos y conozca
cómo implementar los resultados derivados de aplicar esta tecnología
en la actividad y procesos de la compañía.
c) Ingeniero de Datos. Responsable de suministrar o facilitar
eficientemente los datos necesarios a los procesos de aprendizaje.
Algo fundamental en grandes organizaciones.
d) Especialista Regulatorio en Machine Learning. La entrada en vigor del
Reglamento General de Protección de Datos (RGPD) y la
implementación de los bancos de pruebas regulatorios o Sandbox en
España conllevará la necesidad de contar con un experto en
legislación que posea conocimientos sobre el marco regulatorio y
ético.
4. Nuevas regulaciones en el sector financiero. El Machine Learning continuará
siendo especialmente relevante en el sector financiero en 2019. La aplicación
de nuevas normativas en la regulación bancaria, como la PSD2, obligará
indirectamente a estos operadores a ir un paso por delante en el tratamiento
de datos, antes de que la competencia lo haga por ellos.
5. Aplicación de la sensorización e IoT en el sector industrial. Esta es una de
las claves de la futura competitividad de la industria española, que debe
comenzar a incorporar sistemas de sensorización e IoT para el
mantenimiento de los activos industriales y aplicarlos a modelos predictivos
que anticipen el funcionamiento y ciclo de vida de la maquinaria.
6. Creciente importancia de los datos para las Administraciones. Las
instituciones públicas empezarán a sistematizar el gran volumen de datos
que disponen sobre los ciudadanos (comportamiento, tránsito, uso de
espacios públicos…). Se abrirá una brecha entre aquellas administraciones
que tomen decisiones en base a las posibilidades de predicción que ofrece
esta información o aquellas que lo hagan siguiendo el modo tradicional.
Machine learning is undoubtedly a technology that has been impacting exponentially on everyday and business applications. It is our duty as technology leaders to know how to guide business challenges well and provide intelligent solutions. In this document we will take a look at what machine learning is and how we can apply solutions to new business challenges in a completely real world. To internalize and take the first step, it does not require large resources or experts in the field. The trend in recent years, indicates that some companies are committed to Machine Learning (ML) but do so at the moment and largely in a controlled or testing environment and without influencing their main business processes according to various studies, it concludes that making the leap to the application in fundamental business areas is the great challenge that organizations must face in 2019. Although this thesis will be a practical guide and that will allow contextualizing the degree of complexity to initialize, we must always keep in mind and as an objective, the following questions: How to evolve from Machine Learning on a small scale to a larger one? This response has been constructed and generated a kind of trend, which we can summarize in these 6 following contexts: 1. Availability of well-governed data. As Machine Learning progresses, the need for data governance is more understood. Data Governance will gain relevance during 2019 and, thanks to this, professionals may be able to know the sources of data well and identify those that are of quality and those that are not. Learning machines are based on good, unique and reliable training data, so it is essential to apply a good data governance policy. 2. Training of professional organizations to truly understand the possibilities of Machine Learning. This not only implies hiring specialized profiles, but also involves training workers who know the business to understand the possibilities and limitations offered by Machine Learning techniques. 3. Development of new data specialist profiles. In addition to the previous point, the extension of the scope of Machine Learning, will force companies to integrate into their teams new highly specialized profiles, not only that of Data Scientist, but also: a) Business partner with knowledge of Machine Learning. This is a professional who comes from the areas of management of a company and makes strategic or tacit decisions and, in addition, has sufficient analytical training to serve as a link between data scientists and business areas. b) Computer scientist who knows how to integrate algorithms and ML models. A worker who understands how algorithms work and knows how to implement the results derived from applying this technology in the company's activity and processes. c) Data Engineer. Responsible for providing or efficiently providing the necessary data to the learning processes. Something fundamental in large organizations. d) Regulatory Specialist in Machine Learning. The entry into force of the General Data Protection Regulation (GDPR) and the implementation of regulatory or Sandbox test benches in Spain will entail the need for a legal expert who has knowledge of the regulatory and ethical framework. 4. New regulations in the financial sector. Machine Learning will continue to be especially relevant in the financial sector in 2019. The application of new regulations in banking regulation, such as PSD2, will indirectly force these operators to go one step ahead in data processing, before the competition Do it for them. 5. Application of sensorization and IoT in the industrial sector. This is one of the keys to the future competitiveness of Spanish industry, which must begin to incorporate sensorization and IoT systems for the maintenance of industrial assets and apply them to predictive models that anticipate the operation and life cycle of machinery. 6. Increasing importance of data for Administrations. Public institutions will begin to systematize the large volume of data they have about citizens (behavior, traffic, use of public spaces ...). A gap will be opened between those administrations that make decisions based on the prediction possibilities offered by this information or those that do so in the traditional way.
Machine learning is undoubtedly a technology that has been impacting exponentially on everyday and business applications. It is our duty as technology leaders to know how to guide business challenges well and provide intelligent solutions. In this document we will take a look at what machine learning is and how we can apply solutions to new business challenges in a completely real world. To internalize and take the first step, it does not require large resources or experts in the field. The trend in recent years, indicates that some companies are committed to Machine Learning (ML) but do so at the moment and largely in a controlled or testing environment and without influencing their main business processes according to various studies, it concludes that making the leap to the application in fundamental business areas is the great challenge that organizations must face in 2019. Although this thesis will be a practical guide and that will allow contextualizing the degree of complexity to initialize, we must always keep in mind and as an objective, the following questions: How to evolve from Machine Learning on a small scale to a larger one? This response has been constructed and generated a kind of trend, which we can summarize in these 6 following contexts: 1. Availability of well-governed data. As Machine Learning progresses, the need for data governance is more understood. Data Governance will gain relevance during 2019 and, thanks to this, professionals may be able to know the sources of data well and identify those that are of quality and those that are not. Learning machines are based on good, unique and reliable training data, so it is essential to apply a good data governance policy. 2. Training of professional organizations to truly understand the possibilities of Machine Learning. This not only implies hiring specialized profiles, but also involves training workers who know the business to understand the possibilities and limitations offered by Machine Learning techniques. 3. Development of new data specialist profiles. In addition to the previous point, the extension of the scope of Machine Learning, will force companies to integrate into their teams new highly specialized profiles, not only that of Data Scientist, but also: a) Business partner with knowledge of Machine Learning. This is a professional who comes from the areas of management of a company and makes strategic or tacit decisions and, in addition, has sufficient analytical training to serve as a link between data scientists and business areas. b) Computer scientist who knows how to integrate algorithms and ML models. A worker who understands how algorithms work and knows how to implement the results derived from applying this technology in the company's activity and processes. c) Data Engineer. Responsible for providing or efficiently providing the necessary data to the learning processes. Something fundamental in large organizations. d) Regulatory Specialist in Machine Learning. The entry into force of the General Data Protection Regulation (GDPR) and the implementation of regulatory or Sandbox test benches in Spain will entail the need for a legal expert who has knowledge of the regulatory and ethical framework. 4. New regulations in the financial sector. Machine Learning will continue to be especially relevant in the financial sector in 2019. The application of new regulations in banking regulation, such as PSD2, will indirectly force these operators to go one step ahead in data processing, before the competition Do it for them. 5. Application of sensorization and IoT in the industrial sector. This is one of the keys to the future competitiveness of Spanish industry, which must begin to incorporate sensorization and IoT systems for the maintenance of industrial assets and apply them to predictive models that anticipate the operation and life cycle of machinery. 6. Increasing importance of data for Administrations. Public institutions will begin to systematize the large volume of data they have about citizens (behavior, traffic, use of public spaces ...). A gap will be opened between those administrations that make decisions based on the prediction possibilities offered by this information or those that do so in the traditional way.
Description
Keywords
INTELIGENCIA ARTIFICIAL, APRENDIZAJE DE MÁQUINA, APRENDIZAJE AUTOMÁTICO, DEEP LEARNING, TECNOLOGÍA, TRANSFORMACIÓN DIGITAL