We present different topics of our research every Wednesday at 11:45 am in room 105, Obermarkt 17. For news about the EAD-Lunch talks and seminars please feel free to subscribe to EAD-Public@googlegroups.com. (Register here: https://groups.google.com/forum/?hl=de&fromgroups#!forum/ead-public)
Die automatische Verarbeitung von Dokumenten als Vektoren ermöglicht einfache und effiziente Vergleiche von Texten nach gemeinsamen Inhalten. Von Suchanfragen bis hin zur Automatischen Erstellung von domänenspeziefischer Fachliteratur hat sich das Vektorraummodell etabliert und soll kurz vor dem Hintergrund der Information Retrieval dargestellt werden.
The flow shop scheduling problem is one of the many standard scheduling problems faced by the manufacturing industry. The problem comprises of scheduling several jobs over multiple machines without preemption. In this talk, we discuss an extension of the FSP with additional constraint of scheduling all the jobs to reduce the total earliness/tardiness, after the jobs have been processed by each machine. A polynomial algorithm will then be presented which optimizes the earliness/tardiness penalty, for a given job sequence, followed by some preliminary results.
This talk will discuss the implementation of both genetic and simulated annealing algorithms on a GPU in order to solve the Flexible Job-Shop Scheduling Problem (FJSP). This talk will discuss speedups achieved through parallel implementation and discuss methods of further fine-tuning the accuracy of results through adjustment of simulated annealing parameters. The talk will end on approaching the problem by utilizing a hybrid algorithm through combination of both simulated annealing and local search to reach optimal solutions at an accelerated rate.
I will discuss some of the basic concepts surrounding microcontrollers and discuss how they can be simply implemented to perform a large variety of tasks ranging from remote control devices to embedded system monitoring platforms. I will specifically address key topics such as communication protocols, memory architecture, analog-digital conversion, and low-power applications.
Privacy and security concerns can prevent sharing of data and derailing data mining projects. Distributed knowledge discovery, if done correctly, can alleviate these problems. The key is to obtain valid results, while providing guarantees on (non)disclosure of the data. In this talk a back-propagation neural network based methodology is presented, when different sites own rows over the same attributes. Each site learns the classification, but nothing about the data at other sites.
Privacy and security concerns can prevent sharing of data and derailing data mining projects. Distributed knowledge discovery, if done correctly, can alleviate these problems. The key is to obtain valid results, while providing guarantees on (non)disclosure of the data. In this talk a k-means clustering based methodology is presented, when different sites contain different attributes for a common set of entities. Each site learns the cluster of each entity, but nothing about the attributes at other sites.
Smartmetering ist auf dem Vormarsch, doch sind die heutigen Datenlogger nicht unter 2000€ zu bekommen. Nun soll eine preiswertere Alternative gefunden werden. Hauptaugenmerk dabei liegt auf dem M-Bus, doch sollen die gängigen Bussysteme unterstützt werden. Vergleich der Lösung von SBS Lösungsansatz und Raspberry Pi.
The Unrestricted CDD with controllable processing times problem consists of scheduling jobs with controllable processing times on a single machine against a common due-date to minimize the overall earliness/tardiness and the compression penalties of the jobs. The objective of the problem is to find the processing sequence of jobs, the optimal reduction in the processing times of the jobs and their completion times. In this work, we present and prove an essential property for the controllable processing time CDD problem for the unrestricted case along with an exact linear algorithm for optimizing a given job sequence for a single machine with a run-time complexity of O(n), where n is the number of jobs. Henceforth, we implement our polynomial algorithm in conjunction with a modified Simulated Annealing (SA) algorithm and Threshold Accepting (TA) to obtain the optimal/best processing sequence while comparing the two heuristic approaches.
Quality of life in advanced age is closely related to the ability to act autonomously. Due to the needs and preferences of old people, technical assistance systems may support the conditions of a high-level quality of life. The sociological part of VATI – which is an interdisciplinary project between informatics and social sciences – aims to collect information about the needs, preferences and limits of old people in their environment. A longitudinal dataset is the objective and will work as a source on which the VATI-Technology-Navigator will be developed – the so called AAL-Panel. Besides, the navigator will be evaluated by the users in an independent survey – the so called VATI-Panel.
Mind maps have been popularized by Tony Buzan in the 1970s as a technique to organize information around a specific topic. He promoted his conception of radial tree, diagramming key words in a colorful, radiant, tree-like structure, arguing that readers tend to scan the entire page of a document or article in a non-linear fashion, rather than scanning from left to right and top to bottom as "traditional" outlines force them to do. Several studies on mind-mapping report positive effects through using them, like getting a better understanding of concepts and ideas in science and an improved learning/study efficiency compared to conventional note-taking. Mind mapping software can be used to organize large amounts of information, e.g. the amount of information and the existing literature you are confronted with when writing a paper or just preparing a presentation. The purpose of this talk is to give a brief overview of existing mind mapping software as well as clarifying if and how they can be efficiently utilized for organizing literature, creating presentations or even writing a new paper. The ideas pointed out during the talk will be discussed afterwards.
The ability to store, analyze and monitor time-dependent data is important in various fields of Science and Economics. For this purpose, independent of the current field of application and the specific use case, mostly a common structure, a time series, is used. A time series is not a series in a mathematical sense, but a sequence of time dependent data points. In the last couple of years a lot of scientific literature from various fields was published, thematising time series analysis, anomaly detections and trend predictions. Time Series has lead to enhancements in the field of Big Data and its applications. Nevertheless, only a few software solutions, such as Apache Spark, a cluster computing framework and pandas, a library for Python, can take advantage of all that research work. Especially Java developers still have to implement almost everything on their own, as there is no useful framework available. This is, where TSx4J ( a framework for Java) comes in. It is designed as a general purpose framework for time series, easy to integrate and extend, in order to allow you to do whatever you would like to do with time series in Java. The talk will not only cover its main goals and features, but also options for future work and licenses.
Natural Language Processing is currently becoming an increasingly important topic in many directions of CS. The talk introduces basic concepts, frameworks and implementation considerations.