Comunicaciones | Romy R. Ravines

Leveraging Data AI for Cyber Risk Management

Mon, 31 Mar 2025 00:00:00 +0000

Leveraging Data AI for Cyber Risk Management

Keywords: Cyber risk, AI, Data Science

This presentation explores how data science and AI are transforming cyber risk management, especially in industrial and critical infrastructure environments. It highlights the limitations of traditional cybersecurity investments, which often lack financial justification, and introduces a data-driven, risk-based approach that quantifies cyber threats in business terms. By leveraging internal telemetry, external threat intelligence, and advanced modeling techniques—including AI, Bayesian inference, and stochastic simulations—organizations can estimate the financial impact of cyberattacks, prioritize mitigation efforts, and optimize cyber insurance strategies.

The deck also showcases how vulnerabilities can be translated into dollars at risk, enabling smarter decisions about which threats to address first. Through real-world case studies and simulations, it demonstrates how organizations can reduce exposure, justify cybersecurity budgets, and align technical controls with business outcomes. Ultimately, it presents a practical framework for turning cybersecurity from a guessing game into a measurable, strategic business function.

Talk delivered during the III Semana de la Ciberseguridad. The event took place on March 31, 2025, at the Business and Tech Faculty of UAX in Madrid.

You don't have a PDF plugin, but you can download the PDF file.

Cybersecurity Incident Compendium

Wed, 09 Oct 2024 00:00:00 +0000

UAX MAker: Cybersecurity Incident Compendium (CIC)

Keywords: Cyber risk, AI, Data Science, Cyber Incidents, UAX

The second part of the presentation introduces the Cybersecurity Incident Compendium (CIC), a dynamic and enriched knowledge base focused on industrial cyber incidents. It aggregates publicly available data from sources like SEC EDGAR and other commercial databases, addressing the challenges of inconsistency, unstructured formats, and limited visibility. The compendium uses advanced data analytics, machine learning, and AI to structure and enrich incident data, enabling deeper insights into attack patterns, victim profiles, and evolving threats.

The CIC follows a data science life cycle—from weekly data collection and enrichment to categorization and dashboard visualization. It applies techniques like Named Entity Recognition and text similarity to analyze over 20,000 incidents, uncovering hidden patterns such as organizational attractiveness to attackers. The platform helps users understand who is being exploited, the impacts and claims involved, and how threats evolve across sectors and regions. Ultimately, it supports proactive cyber risk management by forecasting expected attacks and identifying key risk drivers.

Talk delivered during the UAX-DeNexus Maker 2024. The event took place on October 09, 2024, at the Business and Tech Faculty of UAX in Madrid.

You don't have a PDF plugin, but you can download the PDF file.

Modeling of Catastrophic Cyber Events in Industrial Environments.

Fri, 10 Mar 2023 00:00:00 +0000

Modeling of Catastrophic Cyber Events in Industrial Environments

Keywords: Cyber risk, AI, Data Science

he presentation focuses on modeling catastrophic cyber events in industrial environments and their impact on portfolio risk accumulation. It introduces the concept of Cyber CAT models, which are designed to assess rare but high-impact cyber incidents—similar to how natural catastrophe models work for events like hurricanes or earthquakes. However, cyber catastrophes pose unique challenges due to their low frequency, lack of historical data, complex correlations, and diverse manifestations of loss across interconnected systems.

The presentation explains how DeNexus’ platform, DeRISK, uses advanced data science to quantify cyber risk by combining inside-out telemetry from operational technology (OT) networks with outside-in threat intelligence. It models attack paths, loss event impacts, and mitigation strategies using statistical methods and simulations. The system supports risk-informed decision-making by estimating financial losses, identifying vulnerabilities, and recommending optimal mitigation actions based on ROI and implementation feasibility. Ultimately, it enables organizations and insurers to understand, manage, and transfer cyber risk more effectively across portfolios.

Masterclass delivered at the Master in Statistics for Data Science, by invitation of the Carlos III University of Madrid.

You don't have a PDF plugin, but you can download the PDF file.

Modeling of Catastrophic Cyber Events in Industrial Environments

Thu, 29 Sep 2022 00:00:00 +0000

Modeling of Catastrophic Cyber Events in Industrial Environments

Keywords: Cyber risk, Monte Carlo Simulation, Loss Distribution

Cyberthreats, as well as the cybersecurity industry, have grown rapidly hand in hand. When you consider that the entire industrial world is dependent on information technology and operational technology solutions, it is not a challenge to understand why. Ransomware targeting critical infrastructure, covert spyware or network failures caused by human error all have a serious impact on an industrial enterprise’s bottom line at various time horizons. This is the definition of cyber risk and defending against it is big business. But what good is a defense without the knowledge of the risk event’s impact? Do we know how cyber risks will affect our operation? How much risk exposure is our business holding at any one time?

Quantifying cyber risk is hard. Cyber risk is dynamic, volatile, and contrary to other sources of risk, cyber risk is subject to human factors ranging, such as motive. Without understanding cyber risk in terms of probabilities and financial terms, no organization can truly mitigate the impact to their business that emerging cyberthreats can pose to their bottom-line, to their employees and shareholders, or even regional laws. Therefore, true cyber risk quantification with real-time evidence-based data is needed. Just as information technology and risk modeling have armed the economy with new measures of efficiency, they should also be used to manage cyber risk.

Charla impartida en el evento 19th International Industrial Cybersecurity Experiences Congress. Europe (Virtual / Face-to-face). Organizado por Centro de Ciberseguridad Industrial de Madrid, España.

https://www.cci-es.org/activities/cciconeuropa2022/

El evento se desarrolló del 29 de Septiembre del 2022.

You don't have a PDF plugin, but you can download the PDF file.

CYBER RISK – FROM THE CONTROL ROOM TO THE BOARD ROOM

Wed, 16 Feb 2022 00:00:00 +0000

CYBER RISK – From the CONTROL room to the BOARD room

Keywords: Cyber risk, Bayesian Inference, Monte Carlo Simulation, Loss Distribution

Quantifying cyber risk is hard. Cyber risk is dynamic, volatile, and contrary to other sources of risk, cyber risk is subject to human factors ranging, such as motive. Without understanding cyber risk in terms of probabilities and financial terms, no organization can truly mitigate the impact to their business that emerging cyber threats can pose to their bottom-line, to their employees and shareholders, or even regional laws. Therefore, true cyber risk quantification with real-time evidence-based data is needed. Just as information technology and risk modeling have armed the economy with new measures of efficiency, they should also be used to manage cyber risk.

Charla impartida en el evento HackOn 2022. Organizado por la Universidad Rey Juan Carlos de Madrid, España.

El evento se desarrolló del 16 de Febrero del 2022.

You don't have a PDF plugin, but you can download the PDF file.

Bayesian Topic Modelling

Sat, 05 Dec 2020 00:00:00 +0000

Bayesian Topic Modeling: Clustering for High-Dimensional Categorical Data applied to Text Analytics

Keywords: Bayesian, Clustering, Variational Inference, Text Analytics, Topic Modeling, MixDir, R

En el contexto empresarial, entender el feedback de los clientes en encuestas, identificar reclamaciones entre muchos e-mails, identificar intervenciones urgentes en informes de mantenimiento, monitorizar el riesgo reputacional en redes sociales e identificar oportunidades de negocio son algunos de los casos donde la utilización del Topic Modeling está siendo un factor diferencial.

La analítica de textos permite encontrar ideas, conceptos y relaciones que se encuentran ‘‘escondidas’’ entre las palabras. Topic Modeling es una forma concreta de hacer analítica de textos, donde el objetivo es agrupar documentos en base a los temas que tratan. En otras palabras, consiste en realizar un análisis de conglomerados en datos tipo texto y la necesidad de trabajar con matrices de alta dimensión - donde se representan las probabilidades de ocurrencias de las palabras en los textos.

En esta charla presentamos la aplicación de un modelo no paramétrico con enfoque bayesiano a la clasificación de una colección de documentos. Utilizamos inferencia variacional porque permite aproximar la distribución a posteriori con menor coste computacional que el requerido por un algoritmo MCMC. Este enfoque, presentado en Dunson and Xing [2009] y Ahlmann-Eltze and Yau [2018], ha sido utilizado en la agrupación de datos categóricos de alta dimensión en campos diferentes a la analítica de textos. Comparamos los resultados con los obtenidos los algoritmos de Machine Learning comúnmente utilizados en esta área, comentamos posibles mejores del modelo y reflexionamos sobre la extensión del uso de este tipo en enfoque en el entorno empresarial.

Charla impartida en la III CONFERENCIA INTERNACIONAL DE PROCESOS ESTOCÁSTICOS, FENÓMENOS ALEATORIOS Y SUS APLICACIONES. Organizada por la Universidad Nacional de Ingeniría (UNI) de Lima Perú.

El evento se desarrolló del 03 – 05 de Diciembre del 2020.

Business and Data Science, Challenges ahead

Fri, 21 Feb 2020 00:00:00 +0000

Masterclass representado a Innova-tsn, por invitación de la Universidad Carlos III de Madrid.

You don't have a PDF plugin, but you can download the PDF file.

Analítica Predictiva y el Teorema de Bayes

Fri, 31 May 2019 00:00:00 +0000

Masterclass representado a Innova-tsn, por invitación de la Universidad Rey Juan Carlos de Madrid.

Anticipar correctamente la demanda de productos o servicios constituye, sin lugar a dudas, un importante ingrediente para la competitividad de las organizaciones. Contar con previsiones de demanda precisas ayuda a tomar decisiones sobre temas y acciones relacionadas con marketing, promociones, inventario, distribución, gestión de categorías, inversión en tiendas, etc. Ahora que vivimos la cultura del Big Data, que estamos inmersos en la transformación digital y que cada día hablamos de la democratización del Machine Learning y la Inteligencia Artificial, todos los focos de atención parecen estar orientados a la explotación de datos y a las decisiones en tiempo real. ¿Qué ha pasado entonces con el impacto de este contexto sobre las “antiguas” necesidades de la empresas?¿Hemos mejorado significativamente los sistemas de previsión de demanda?

En esta Masterclass abordaremos, desde un punto de vista práctico, aspectos claves para acertar en la elección del Sistema de Analítica Predictiva de una organización. Modelos y algoritmos a utilizar, diagnosis y explicatividad de resultados, uso del conocimiento de negocio vía inferencia bayesiana, son algunos de los elementos que las empresas empiezan a requerir y que señalan el camino futuro de las decisiones informadas basadas en la ciencia de datos.

You don't have a PDF plugin, but you can download the PDF file.

Statistics for Data Science from the business perspective in 2019

Fri, 01 Mar 2019 00:00:00 +0000

Masterclass representado a Innova-tsn, por invitación de la Universidad Carlos III de Madrid.

You don't have a PDF plugin, but you can download the PDF file.

Conociendo mejor la Experiencia de Cliente mediante Text Analytics

Thu, 21 Feb 2019 00:00:00 +0000

Ponencia representado a Innova-tsn, uno de los patrocinadores del evento.

Ver transmisión via Twitter

You don't have a PDF plugin, but you can download the PDF file.

Marketing y Analítica Avanzada

Sat, 30 Jun 2018 00:00:00 +0000

Masterclass representado a Innova-tsn, por invitación de la Universidad Rey Juan Carlos de Madrid.

You don't have a PDF plugin, but you can download the PDF file.

In-Scoring, Machine Learning y Big Data al servicio de la Puntuación Inteligente

Wed, 06 Jun 2018 00:00:00 +0000

Ponencia representado a Innova-tsn, uno de los patrocinadores del evento.

Ver entrevista posterior a la ponencia

You don't have a PDF plugin, but you can download the PDF file.

¿Qué es Big Data? Le preguntamos a Twitter

Wed, 14 Mar 2018 00:00:00 +0000

Ponencia representado a Innova-tsn, uno de los patrocinadores del evento.

Ver Video de la ponencia

Ver Entrevista posterior a la ponencia

Modelos Dinámicos y Bayesianos para Marketing

Fri, 09 Jun 2017 00:00:00 +0000

Masterclass representado a Bayes Forecast, por invitación de la Universidad Rey Juan Carlos de Madrid.

You don't have a PDF plugin, but you can download the PDF file.

Modelos de Previsión (Bayesianos) en la Empresa

Thu, 30 Jun 2016 00:00:00 +0000

Masterclass representado a Bayes Forecast, por invitación de la Universidad Rey Juan Carlos de Madrid.

You don't have a PDF plugin, but you can download the PDF file.

Previsión de Audiencia. Aplicación de Modelos Jerárquicos Bayesianos

Wed, 26 Nov 2008 00:00:00 +0000

Charla dentro del Ciclo de Seminarios del Departamento de Estadística e Investigación Operativa (DEIO) de la Universidad Rey Juan Carlos de Madrid (URJC).

You don't have a PDF plugin, but you can download the PDF file.

Marco Conceptual en Indicadores de Pobreza

Thu, 17 Apr 2008 00:00:00 +0000

Charla dentro del Observatorio de Equidad en Salud y Foro permanente para la propuesta ciudadana y Toma de Decisiones basadas en evidencias, organizado por el Instituto Nacional de Estadística e Informática de Perú (INEI) y USAID, Perú.

You don't have a PDF plugin, but you can download the PDF file.

Conversatorio sobre construcción del Mapa de Pobreza

Tue, 18 Dec 2007 00:00:00 +0000

Charla con Consultores del Instituto Nacional de Estadística e Informática de Perú (INEI), organizado por el Centro de Investigación y Desarrollo (CIDE) y financiado por el Banco Interamericano de Desarrollo (BID)

You don't have a PDF plugin, but you can download the PDF file.

A Bayesian Approach for the Rainfall-Runoff Problem in Multiple Basins

Thu, 01 Feb 2007 00:00:00 +0000

X Congreso latinoamericano de probabilidad y estadística matemática. CLAPEM. Lima-Perú.

We present an analysis of runoff and rainfall data from Rio Grande, a basin located in the northeast of Brazil. The main problems we face here are:

to model runoff and rainfall jointly, taking into account their different spatial units,
to use stochastic models where all the parameters have physical interpretations, and (iii) to model multiple runoff series simultaneously.

The intrinsically uncertain nature of those hydrological processes makes Bayesian analysis natural within this field. Our approach is based in dynamic models. For run-off, we use transfer functions, and for rainfall, we use a change of support from a multisite spatial model. Besides the computational effort to implement the proposed models, it is worth pointing out that some methodological novelties are also implemented.

The data consist of monthly recorded series from January 1984 to September 2004, at three runoff stations and nine rainfall stations irregularly located in an area of drainage of 37522.48 km2. Model assessment, spatial interpolation and temporal predictions were part of our analysis. Results has shown that our approach is a promising tool for the runoff-rainfall analysis.

You don't have a PDF plugin, but you can download the PDF file.

An efficient sampling scheme for dynamic generalized models

Fri, 02 Jun 2006 00:00:00 +0000

Eigth Valencia International Meeting, Spain.

We consider non-linear state-space models for univariate time series with non-Gaussian observations in a Bayesian framework. For practitioners computational efficiency is always a desirable feature. We present a scheme for multimove sampling of the states parameters of such models within a Gibbs sampling algorithm. Our proposed sampling scheme combines the conjugate updating of West, Harrison, and Migon (1985) for dynamic models in the exponential family, with the Backward Sampling of Frühwirth-Schnater (1994).

We call this algorithm Conjugate Updating Backward Sampling (CUBS).

We notice that CUBS significantly reduces the computing time needed to attain convergence of the chains, being also simple to implement. Samples from the posterior distribution of other parameters in the models can be obtained using standard MCMC techniques. A quite extensive Monte Carlo study was conducted in order to compare the results obtained with CUBS with those obtained from some algorithms previously proposed in the Bayesian literature.

You don't have a PDF plugin, but you can download the PDF file.

A Bayesian approach for the rainfall-runoff problem. The case of Rio Grande Basin

Mon, 03 Apr 2006 00:00:00 +0000

Charla dentro del Ciclo de Seminarios del Departamento de Estadística e Investigación Operativa (DEIO) de la Universidad Rey Juan Carlos de Madrid (URJC), como parte de la Estancia Pre-Doctoral financiada por CNPq (Brasil).

You don't have a PDF plugin, but you can download the PDF file.

Revisiting Distributed Lag Models Through a Bayesian Perspective

Mon, 01 Dec 2003 00:00:00 +0000

You don't have a PDF plugin, but you can download the PDF file.

Inferência em Modelos Hierárquicos Generalizados sob Planos Amostrais Informativos

Wed, 01 Jan 2003 00:00:00 +0000

Trabajo presentado en:

VIII Escola de Modelos de Regressão, EMR. Conservatória, 2003, Brasil.
XVI Simpósio Nacional de Probabilidade e Estatística, 2004, Caxambu, Brasil

You don't have a PDF plugin, but you can download the PDF file.

Métodos Estadísticos Aplicados con la Enaho

Tue, 28 Nov 2000 00:00:00 +0000

Charla impartida en el VI Coloquio Nacional de Estadística, organizado por la Universidad Nacional de Ingenieria (UNI), representando al Instituto Nacional de Estadística e Informática (INEI) de Perú.

You don't have a PDF plugin, but you can download the PDF file.