Auflistung nach:
Auflistung S17 - SKILL 2021 - Studierendenkonferenz Informatik nach Erscheinungsdatum
1 - 10 von 16
Treffer pro Seite
Sortieroptionen
- TextdokumentAnomaly Detection in Motion Timeseries using the Bosch XDK and Dynamic Time Warping(SKILL 2021, 2021) Mejía, Julián Rico; Isaías, Oscar Aguilar Aguila; Paschapur, PriyankaThis paper presents the development of an anomaly detector for robotic movements using the dynamic time warping (DTW) algorithm and its implementation in Matlab. Data was collected by mounting the Bosch Cross-Domain Development Kit (XDK) sensor on a collaborative robot arm (Cobot), aiming at industrial applications in need for motion anomaly detection during repetitive tasks. The paper discusses practical issues like parameter tuning as well as algorithmic variants such as decoupling accelerometer and gyroscope data.
- TextdokumentData-based Transparency and Leadership in Small and Medium-sized Enterprises(SKILL 2021, 2021) Mayer, CarmenBased on the increasing usage of Information Systems (IS), the amount of employee-specific data in companies is rising. As this data is more often used for leadership, referred to as data-based leadership, the question about complete transparency and its consequences in companies needs consideration. This work therefore, aims to analyze the amount of gathered employee-specific data, the resulting data-based leadership, and the exercised control and transparency in small and medium enterprises (SME) which have limited experience in using digital leadership approaches. The applied case study provides qualitative insights into these aspects. This case study is based on five selected SMEs from different industries. With my study, I enhance control theory and derive practical recommendations for a sustainable handling of employee-data for leadership.
- TextdokumentDesigning an ethical technology project with the help of Data Feminism(SKILL 2021, 2021) Gleißner, Lea-Kathrin; Bui, Magdalena; Kühn, Fey; Nenninger, AmelieAlgorithms and new technologies help people in several life situations, but society pays a high price for their advantages. Several scandals occurred recently, showing that algorithms are neither neutral nor fair – quite the contrary: They discriminate people as humans do. One approach to create less biased data science projects is the “Data Feminism” method, presented by Catherine D’Ignazio and Lauren F. Klein in their book of the same title. This paper evaluates how feasible the method can be implemented in student projects based on the experiences four Leipzig students made by trying to implement the method into their project ‘Questioning Street Names Leipzig’. The paper focusses on three main concepts: subjective viewpoints and context, crediting all forms of labour, and building and linking communities through public tagging events, thus opening the academic question for some citizen science help. The project utilizes open data and open data sources such as Wikidata and OpenStreetMap. The authors of “Data Feminism” want to encourage students, as well as academic professionals, to think about their bias in their data and to use the data feminism approach to reduce the impact of them and create more ethical computer science projects.
- TextdokumentMultiple Sequence Alignment using Deep Reinforcement Learning(SKILL 2021, 2021) Joeres, RomanMultiple sequence alignment (MSA) is one of the primal problems in biology and bioinformatics. The question of how to align multiple sequences correctly is crucial for many other fields of research, e.g., gaining information about the evolutionary distance of two or more sequences and therefore about their corresponding species, finding protein targets for drugs, or finding a drug for a certain target protein. Reinforcement learning (RL), and especially deep reinforcement learning (DRL), has become popular in recent years. To name just a few, DRL has shown major success in complex games such as Atari Games, Chess, and Go. We model the problem of aligning multiple sequences as a Markov decision process (MDP) and examine the performance of different (D)RL algorithms compared to state-of-the-art tools.
- TextdokumentBicycle Detection from Top View Perspective in Surveillance System using Convolutional Neural Network(SKILL 2021, 2021) Ramkumar, Sanal DarshidBicycle detection and tracking from top view perspective using deep learning is a highly active research area for video surveillance and automatic ticket generation in Advanced Public Transportation System (APTS). People detection using conventional cameras has received massive attention for video surveillance inside public transportation systems but inattentive towards bicycle detection. Experimentation is performed on You Only Look Once (YOLO), Faster Regional-Convolutional Neural Network (Faster R-CNN) and Single Shot Multibox Detector (SSD). Due to the sparse availability of dataset for this work, a customized dataset was recorded in the Media Computing lab, Junior Professorship of Media Computing, TU Chemnitz, Germany. The customized dataset was recorded using a wide-angle smart stereo sensor (S2000, Intenta GmbH) mounted in bird’s eye perspective. Furthermore, two additional datasets were recorded using a mobile camera representing indoor and outdoor bicycle parking area. This paper provides best case solution for bicycle detection from a top view perspective.
- TextdokumentThe Impact of Domain Knowledge on Applying Machine Learning Methods to Exoplanet Detection(SKILL 2021, 2021) Nguyen, The-Gia LeoExoplanets do not emit electromagnetic waves which makes it challenging to detect them. Based on transit photometry, we trained a neural network on NASA Kepler space telescope data to detect exoplanets based on light intensity curves. We showcase, that with a well designed data pipeline, a small neural network is sufficient to achieve state-of-the-art performance, saving both computation time and hardware cost. The strongest improvement in performance could only be achieved by adding domain specific processing steps to the data pipeline. Domain knowledge was essential in selecting the appropriate machine learning concepts that are beneficial to solving the problem and have a higher impact on the performance than the actual classification method itself. We encourage to consider the data pipeline as an additional component, besides the classification model, that can potentially improve the overall performance.
- TextdokumentSKILL 2021 - Studierendenkonferenz Informatik - Komplettband(SKILL 2021, 2021) Geselllschaft für Informatik e.V.
- TextdokumentExplainable Diagnosis of COVID-19 from Chest X-ray Images via CNNs(SKILL 2021, 2021) Arkan, Emre; Beckert, Jan MalteThis work demonstrates how Convolutional Neural Networks ( CNN s) can be used to identify signs of COVID-19 from Chest X-rays (CXR s) and discusses the challenges of deep learning with small datasets. In order to validate the model’s performance, two novel explanation methods LIME and Grad-CAM are explored. Additionally, they serve to further increase users’ confidence in specific classifications. Since the explanation results revealed model biases, additional preprocessing mechanisms were explored: A U-Net-based lung segmenter is introduced to the preprocessing pipeline, which masks all non-lung parts of the CXRs images. Subsequently, the segmentation and non-segmentation results were evaluated with regard to both their performance metrics and interpreted explanation results.
- TextdokumentBuilding a GAN for Replicating Epithelial Impedance Spectra for ML-based Pattern Recognition(SKILL 2021, 2021) Jurkschat, Lena; Schindler, BenjaminImpedance spectroscopy is a common method in the field of biotechnology to measure electrical conductivity of special cell lines (i.e. ephitelial). Based on the measured impedance spectra, machine learning (ML) techniques including random forests and feedforward networks are increasingly used to determine physiological properties of the underlying cell tissue and to detect a wide range of diseases. However, training ML models for this purpose typically requires large amounts of data and real cell tissue measurements are costly to obtain due to their experimental setup. This paper introduces a Generative Adversarial Network (GAN) which meets the high demand for training data by replicating impedance spectra from a given data set. As a proof of concept, we show that GANs are capable of generating spectra that have a similar shape to the original ones and could therefore be used to overcome a lack of training data.
- TextdokumentConcolic-Fuzzing of JavaScript Programs using GraalVM and Truffle(SKILL 2021, 2021) Delhougne, RobertThe scripting language JavaScript has established itself as a central component of the modern internet. However, the dynamic execution model of the language limits the support for source-code analysis, which leaves a developer without essential tools to maintain safety and security requirements. This paper describes a concolic-fuzzer based on the GraalVM to automatically test JavaScript programs. The fuzzer shows promising results in both code coverage and runtime evaluations and provides developers with additional features such as special analysis targets.