Hasso-Plattner-Institut20 Jahre HPI
Hasso-Plattner-Institut20 Jahre HPI
  
Login
 

The period Spring 2019 started with the HPI Future SOC Lab Day on April 09, 2019. The ended with the HPI Future SOC Lab Day on November 12, 2019. 

35 research projects have used the Lab's IT infrastructure. If you would like to receive more information about one or more projects, please contact us.

Germany

Deep learning and in-memory text analytics for evidence-based knowledge synthesis

Abstract

IMDB and deep learning based NLP technology shall be used to help processing the vast volume of medical literature and make latest medical knowledge accessible for evidence-based medicine. We try to extract information clinical information from a large corpus of medical documents and link them to recommendations in clinical practice guidelines.

 

Researchers

Principle Investigator: Prof. Dr.-Ing. Bert Arnrich || Contact Author: Florian Borchert || Hasso Plattner Institute, Digital Health Center

BlockBI

Abstract

While blockchain is currently discussed in a number of different scenarios, still a lot of questions remain unanswered. One of those questions is the analysis of data that is stored on the blockchain. The project aims at creating Business Intelligence like reports of  Business Intelligence (BI) of blockchain based data.

 

Researchers

Principle Investigator: Prof. Dr. Marc Jansen || Contact Author: Prof. Dr. Marc Jansen || HS Ruhrwest

Federated Learning Applications in Digital Health

Abstract

I want to investigate how the novel federated learning approach can improve and simplify building machine learning models in a digital health environment.

 

Researchers

Principle Investigator: Prof. Dr. Bert Arnrich || Contact Author: Bjarne Pfitzner || Hasso Plattner Institute, Digital Health Center

Exploring Game-Theoretic Formation of Realistic Networks

Abstract

Many real world networks share the same structural properties. We have developed an agent-based game-theoretic model which promises an explanation of the structure of real world networks. In this project we want use the HPI Future SOC Lab to investigate the properties of large generated networks and to validate them with real-world data.

 

Researchers

Principle Investigator: Prof. Dr. Tobias Friedrich || Contact Author: Dr. Pascal Lenzner || Hasso Plattner Institute

Behaviour-based authentication: feature engineering and evaluation based on large user profiles

Abstract

Our project contributes to the field of behaviour-based authentication. In the last years we collected a great amount of walking sequences of several people. As this dataset is too large to be processed on a normal machine, we hope to evaluate and improve our authentication model by the support of additional resources.

 

Researchers

Principle Investigator: Prof. Dr. Christoph Meinel || Contact Author: Vera Weidmann || neXenio GmbH

Gait Analysis and Activity Recognition using Deep Learning

Abstract

Gait analysis is an important tool to assess diseases or clinical conditions that have motor symptoms, such as Parkinson's disease, stroke or knee replacement surgeries. Currently, most of the gait analysis systems require a larger and expensive clinical setup. The goal of this research project is the development of a mobile gait analysis solution.

 

Researchers

Principle Investigator: Prof. Dr. Bert Arnrich || Contact Author: Justin Albert || Hasso Plattner Institute, Digital Health Center

Integrating Hardware Accelerators in Virtualized Environments

Abstract

In this project, we study mechanisms for integrating hardware accelerators in virtual machines and cloud infrastructures. Exemplary workloads include In-Memory Databases, scientific computation and multimedia applications. This project is a continuation of preceding projects conducted in the Spring and Fall Periods of 2018.

 

Researchers

Principle Investigator: Prof. Dr. Andreas Polze || Contact Author: Max Plauth || Hasso Plattner Institute

Measurement-Based Software Performance Engineering for Microservices and Multi-Core Systems

Abstract

DevOps, microservices, and multi-core systems are gaining considerable attraction in research and practice. We would use the requested HPI Future SOC Lab resources to investigate experimental evaluation of our activities on "DevOps-oriented Load Testing for Microservices" and "Software Performance Engineering for Multi-Core Systems".

 

Researchers

Principle Investigator: Dr.-Ing. Andre van Hoorn || Contact Author: Dr.-Ing. Andre van Hoorn || University of Stuttgart

GPU-Accelerated Causal Inference

Abstract

We want to investigate the use of GPUs for enterprise relevant machine learning algorithms. In particular, we aim to adapt an implementation of an algorithm for causal inference to target the GPU as an execution device. This reduces the execution time of the algorithm and in addition free CPU cycles to be used by other enterprise applications.

 

Researchers

Principle Investigator: Dr. Matthias Uflacker || Contact Author: Christopher Schmidt || Hasso Plattner Institute

An Energy-Aware Runtime System for Heterogeneous Clusters

Abstract

We are planning to evaluate our Albatross runtime system (DOI: 10.1145/3217189.3217193) for energy-efficient and economic processing on heterogeneous compute infrastructures in the HPI Future SOC Lab in cooperation with the OSM Group (HPI).

 

Researchers

Principle Investigator: Prof. Dr.-Ing. habil Wolfgang Schröder-Preikschat || Contact Author: Dr.-Ing. Timo Hönig || Friedrich-Alexander-Universität Erlangen-Nürnberg

Nursing Activity Recognition

Abstract

Documentation and planning of patients’ treatment are one of the biggest time-consuming activities for nurses, which can be  reduced by using deep learning methods such as Long Short-Term Memory networks. Therefore activities of nurses will be recorded and analyzed using commodity accelerometers on a smartwatch.

 

Researchers

Principle Investigator: Professor Bert Arnrich || Contact Author: Orhan Konak || Hasso Plattner Institute, Digital Health Center

ForME: Forecasting Mental Earthquakes

Abstract

Our goal is to predict psychological crisis (“mental earthquakes”) based on smartphone data. We focus on meta-information about messages, voice quality and breathing. To detect detect critical changes in the individuals’ behavior, we use machine learning algorithms on multimodal time-series to recognized anomalies in such highly complex streams.

 

Researchers

Principle Investigator: Prof. Dr. Erwin Böttinger || Contact Author: Hanna Drimalla || Hasso-Plattner Institute, Digital Health Center

Improving Test Suite Generation by Testing Google Play's Top 1000 Apps III

Abstract

Automated generation of test suites tackles the complexity of testing apps. The selection of the generation method is typically done in a trial-and-error fashion due to unknown characteristics of the apps. Analyzing the fitness landscapes of Google Play's top 1000 apps, we want to understand the characteristics and improve the generation method.

 

Researchers

Principle Investigator: Dr. Thomas Vogel || Contact Author: Dr. Thomas Vogel || Humboldt-Universität zu Berlin

Web-based distribution of large amounts of data in local networks

Abstract

We are developing a peer-to-peer CDN for the school-cloud. A peer-to-peer based CDN is able to use the time and localness of data to reduce the needed bandwidth by transferring data within the local network between two peers. To evaluate this, we want to compare different peer meshing strategies and explore technical limitations.

 

Researchers

Principle Investigator: Dipl. Inf (FH) Jan Renz || Contact Author: Tim Friedrich || Hasso Plattner Institute

Live Migration Optimization using Machine Learning

Abstract

In this research project, we focus on the impact of running live migration process on the datacenter physical resources, build models that relates the migration volume of the VM with the CPU, memory, network and power overhead.

 

Researchers

Principle Investigator: Prof. Dr. Christoph Meinel || Contact Author: Mohamed Elsaid || Hasso Plattner Institute

Towards production-ready tools for self-driving data management with deep reinforcement learning

Abstract

According to practitioners, deep reinforcement learning (DRL) is poised to revolutionize how autonomous systems are built. However there are several challenges for making these tools production-ready. In this project we define 5 sub-projects that seek to address what we identify as the main challenges for self-driving data management DRL solutions.

 

Researchers

Principle Investigator: Prof. Dr. rer. nat. hab. Gunter Saake || Contact Author: Gabriel Campero Durand || University of Magdeburg

Impact of TLB Shootdown Latency

Abstract

On large-scale multicore systems, the latency of TLB shootdowns is dominated by a multicast mechanism based on IPIs. This project evaluates the impact of low latency TLB shotdowns on application performance by replacing the sequential multicast mechanism in the Linux kernel by a tree-based variant with logarithmic scaling.

 

Researchers

Principle Investigator: Prof. Dr.-Ing. Jörg Nolte || Contact Author: Robert Kuban || BTU Cottbus-Senftenberg

Decision Support Systems for Telemedicine

Abstract

The use of telemedical interventional monitoring in heart failure has been shown to reduce the percentage of days lost due to unplanned cardiovascular hospitalisations or all-cause death as well as all-cause and cardiovascular mortality. This project evaluates the use of decision support systems to scale the monitoring.

 

Researchers

Principle Investigator: Prof. Dr. Andreas Polze || Contact Author: Jossekin Beilharz || Hasso Plattner Institute

Europe

Glucose-level prediction from voice

Abstract

We propose a large-scale study on glucose level detection from voice. We will start a distributed application for 200 users in four hospitals. The data will contain voice data collected from patient during 9 days and all the other date the hospitals have concerning the history of their diabetes. Simply please leave our data/machines as they are.

 

Researchers

Principle Investigator: Prof Lars Lundberg || Contact Author: Assit. Prof Julia Sidorova || Blekinge Tekniska Högskola

WEEVIL- Third (WEEVILT)

Abstract

The WEEVILT project is designed as the extension of the previous one (WEEVILS). This document details the main aims, scope and schedule development for the WEEVILT project. This project is a continuation of previous one, that is, this submission is a renewal for accessing HPI Future SOC Lab infrastructure.

 

Researchers

Principle Investigator: Prof. Dr. Carlos Juiz || Contact Author: Belen Bermejo || University of the Balearic Islands

Adversarial Risk Analysis for Obfuscation Attacks

Abstract

Malware detection is still an important issue in the current world due to the huge quantity daily attacks. Obfuscation procedures are used by adversaries to create metamorphic malware difficult to detect. These procedures include dead-code insertion, register reassignment, subroutine reordering, code transportation, among others. 

 

Researchers

Principle Investigator: Professor David Ríos Insua || Contact Author: Alberto Redondo Hernández || ICMAT-CSIC

A Comprehensive Characterization of HTAP – Extension Request

Abstract

The popularity of large-scale real-time analytics applications keeps rising. These applications require distributed data management systems that can handle fast concurrent transactions and analytics on the recent data (i.e., HTAP). The goal of this project is to characterize/analyze several HTAP systems on large multicore hardware and clusters.

 

Researchers

Principle Investigator: Associate Prof. Pinar Tozun || Contact Author: Associate Prof. Pinar Tozun || IT University of Copenhagen

Process-based machine learning method to analyze HEI compliance

Abstract

In the era of fourth revolution educational institutions have a responsibility to educate the future employees. They need predictions of future competences. Our research aims at creating a data warehouse to assess future job competences collecting time series data from job portals and transforms them into the data warehouse developing process.

 

Researchers

Principle Investigator: Associate Prof. Katalin Ternai || Contact Author: Associate Prof. Ildikó Szabó || Corvinus University of Budapest

Deep Representation Learning on Large Attributed Graphs

Abstract

Real-world graphs are often associated with a rich set of attributes (e.g., text, image) that can significantly influence the interactions, such as in political campaigns. This project wants to develop a Deep Learning unsupervised model able to generate a representation that encodes both the relational structure and node attributes in these graphs.

 

Researchers

Principle Investigator: Prof. Vincenzina Messina || Contact Author: PhD Debora Nozza || University of Milano-Bicocca

Text-based Knowledge Graph Embeddings

Abstract

In the following we describe our project we plan to use text to generate embeddings of entities that are describe in a Knowledge Graph. We want to extend our current work by using new text sources to generate the embeddings, considering embedding in hyperbolic spaces and introducing reasoning mechanisms.

 

Researchers

Principle Investigator: Associate Prof. Andrea Maurino || Contact Author: PhD Student Federico Bianchi || University of Milan-Bicocca

Quality Assessment of RDF datasets at Large-Scale

Abstract

Over the last years, Linked Data has grown steadily. Today, we count more than 10,000 datasets being available online following linked data standards. Gratefully, these standards have made machine-readable and interoperable data. We describe the first distributed in-memory approach for computing different quality metrics for RDF datasets.

 

Researchers

Principle Investigator: Dr. Anisa Rula || Contact Author: Dr. Anisa Rula || University of Milano-Bicocca

Improving the Quality of Art Market Data with Machine Learning and Linked Open Data

Abstract

This proposal describes a framework for improving the quality of art market data. The process consists of several steps: data cleansing, data enrichment, and inferring new findings. LOD and deep learning techniques are used for inferring new findings, such as style and object detection in paintings.

 

Researchers

Principle Investigator: Prof. Dr. hab. Witold Abramowicz || Contact Author: Mgr inż. Dominik Filipiak || Poznań University of Economics and Business

Benchmarking Java on Ethernet Cluster

Abstract

The aim of this project is to check the scalability of parallel, network intensive microbenchmarks and application written in Java, using the PCJ library, HPC Challenge 2014 award-winning Java library for high-performance parallel computing, on the 1000 Core Cluster - with high performance Ethernet interfaces.

 

Researchers

Principle Investigator: Dr. Marek Nowicki || Contact Author: Dr. Marek Nowicki || Nicolaus Copernicus University in Toruń

CitySensing

Abstract

Big mobility and IoT data processing and analytics in Smart Cities. The planned activities are focused on research and development of methods, tools and software systems for efficient and effective processing, analysis and mining of Big mobility data collected leveraging mobile crowd sensing and Internet of Things paradigms in Smart Cities.

 

Researchers

Principle Investigator: Prof. Dragan Stojanovic || Contact Author: Prof. Dragan Stojanovic || University of Nis

Worldwide

Overfitting on purpose to design new algorithms

Abstract

The project takes place in the greater area of heuristic optimisation. In 2017, Prof Tobias Friedrich (HPI, Chair for Algorithm Engineering) and I have explored the concept of automated algorithm configuration to design new search operators. The experiments inspired theoretical investigations, which proved that the new methods beat state-of-the-art.

 

Researchers

Principle Investigator: Dr. Markus Wagner || Contact Author: Dr Markus Wagner || University of Adelaide

Comparative of techniques for similarity measures applied for Musical Genre Classification

Abstract

The objective of this project is to make a deep comparative between different techniques for similarity measures, focusing in the Musical Genre Classification.

 

Researchers

Principle Investigator: Joao Sauer || Contact Author: Joao Sauer || Universidade Federal do Paraná

Intelligent Software Development with Deep Learning

Abstract

The overarching goal of my research is to improve software developers’ productivity and software product quality based on the big data of software repositories such as the code in GitHub, Q&A discussions in Stack Overflow. I will further automate GUI development of mobile apps and enhance the robustness of Deep Learning system.

 

Researchers

Principle Investigator: Dr. Chunyang Chen || Contact Author: Dr. Chunyang Chen || Monash University

Scalable Batch and Real-time Analytics of Trajectory Data

Abstract

Large  volume  of  sensor  networks  and  trajectories  of  mobile  objects are  collected.  Such  data  offer  us  high  value  knowledge  to  understand moving objects and locations, fostering a broad range of applications in smart  cities,  enabling  intelligent  transportation  systems  and  intelligent urban computing.

 

Researchers

Principle Investigator: Dr. Rim Moussa || Contact Author: Dr. Rim Moussa || ENI-Carthage

On-Demand Representation Learning for Heterogeneous Data

Abstract

We propose an On-Demand Representation Learning service that flexibly learns suitable vector representations for items submitted by a client, drawing on a combination of several large-scale data sources. This enables the client's machine learning algorithms to better cope with new signals that never occurred in their domain-specific training data.

 

Researchers

Principle Investigator: Prof. Dr.-Ing. Gerard de Melo || Contact Author: Prof. Dr.-Ing. Gerard de Melo || Rutgers University