Data-Driven Future Forum: research trends in data management, data security and data-driven business models

Abstract

Large-scale data analytics and machine learning techniques underpin many modern data-driven applications. The end-to-end lifecycle of these applications poses, however, significant data management challenges, especially at scale. In a first part, we will summarize these data management challenges, provide an overview of existing systems and tools, and discuss requirements for necessary abstractions. In the second part, we will describe SystemML as a representative system for declarative, large-scale machine learning. SystemML provides an R-like syntax and automatically compiles these high-level linear algebra programs into hybrid runtime plans that combine single-node in-memory operations, and distributed operations on Spark. We will motivate declarative ML, provide an up-to-date overview of SystemML, its compiler and runtime, as well as APIs for different deployments. Interestingly, there are many similarities to traditional relational data management and query processing. Finally, we draw several conclusions on the status and future directions of data management for machine learning but also machine learning for data management problems.

Abstract

While the Internet of Things (IoT) enables a huge range of novel applications, also the security challenges arising through the deployment of IoT technologies are unprecedented. In fact, IoT devices are exposed to a wide range of attacks ranging from remote attacks via the network to local physical attacks. At the same time, the value of information that is collected and processed by IoT devices is continuously rising.

This talk provides an overview of IoT security challenges. Besides focusing on classical security questions and countermeasures, the talk in particular also focuses on side-channel attacks, which allow the bypassing of security mechanisms by exploiting properties, such as the timing behavior or the power consumption, of the underlying hardware.

Abstract

Cryptography provides the mathematical building blocks at the very foundation of any secure application. In addition to confidentiality and authenticity of data, novel cryptographic algorithms can address a wide spectrum of security and privacy challenges and enable innovative solutions. However, the advent of quantum computers threatens many of the most widely deployed algorithms. We discuss how quantum computers impact security and how new concepts can protect future applications.

Abstract

Digitization is generating  data in all areas of business. Modern data analytics methods open up these large amounts of data for business value creation.  Data is now a strategic resource for business model development. Data can be utilized in every element of a business to create value: Data can constitute new value propositions, or enrich existing value propositions. It can support value creation processes  (e.g. improving production processes) and value capturing (e.g. selling data or data-generated information). In this talk, Gert Breitfuss (Senior Expert on Data-Driven Business Model Innovation at Know-Center) will talk about the relevance of data in business, challenges associated with transforming and creating data-driven business models, and latest research ongoing at the Know-Center within the research area Data-Driven Businesss.

Abstract

Following the expert talks, Stefanie Lindstaedt will outline the overall picture of the research focus at the Know-Center and explain which future contributions the industry can expect from Austria’s leading Research Center for Data-driven Business and Big Data Analytics.