Logo des Repositoriums

Auflistung nach Schlagwort "Machine learning"

1 - 10 von 47
  • Zeitschriftenartikel
    Admire LVQ—Adaptive Distance Measures in Relevance Learning Vector Quantization
    (KI - Künstliche Intelligenz: Vol. 26, No. 4, 2012) Biehl, Michael
    The extension of Learning Vector Quantization by Matrix Relevance Learning is presented and discussed. The basic concept, essential properties, and several modifications of the scheme are outlined. A particularly successful application in the context of tumor classification highlights the usefulness and interpretability of the method in practical contexts. The development and putting forward of Matrix Relevance Learning Vector Quantization was, to a large extent, pursued in the frame of the project Adaptive Distance Measures in Relevance Learning Vector Quantization—Admire LVQ, funded through the Nederlandse Organisatie voor Wetenschappelijk Onderzoek (NWO) under project code 612.066.620, from 2007 to 2011.
  • Zeitschriftenartikel
    AI Startup Business Models
    (Business & Information Systems Engineering: Vol. 64, No. 1, 2022) Weber, Michael; Beutter, Moritz; Weking, Jörg; Böhm, Markus; Krcmar, Helmut
    We currently observe the rapid emergence of startups that use Artificial Intelligence (AI) as part of their business model. While recent research suggests that AI startups employ novel or different business models, one could argue that AI technology has been used in business models for a long time already—questioning the novelty of those business models. Therefore, this study investigates how AI startup business models potentially differ from common IT-related business models. First, a business model taxonomy of AI startups is developed from a sample of 100 AI startups and four archetypal business model patterns are derived: AI-charged Product/Service Provider, AI Development Facilitator, Data Analytics Provider, and Deep Tech Researcher. Second, drawing on this descriptive analysis, three distinctive aspects of AI startup business models are discussed: (1) new value propositions through AI capabilities, (2) different roles of data for value creation, and (3) the impact of AI technology on the overall business logic. This study contributes to our fundamental understanding of AI startup business models by identifying their key characteristics, common instantiations, and distinctive aspects. Furthermore, this study proposes promising directions for future entrepreneurship research. For practice, the taxonomy and patterns serve as structured tools to support entrepreneurial action.
  • Zeitschriftenartikel
    AI-Enhanced Hybrid Decision Management
    (Business & Information Systems Engineering: Vol. 65, No. 2, 2023) Bork, Dominik; Ali, Syed Juned; Dinev, Georgi Milenov
    The Decision Model and Notation (DMN) modeling language allows the precise specification of business decisions and business rules. DMN is readily understandable by business users involved in decision management. However, as the models get complex, the cognitive abilities of humans threaten manual maintainability and comprehensibility. Proper design of the decision logic thus requires comprehensive automated analysis of e.g., all possible cases the decision shall cover; correlations between inputs and outputs; and the importance of inputs for deriving the output. In the paper, the authors explore the mutual benefits of combining human-driven DMN decision modeling with the computational power of Artificial Intelligence for DMN model analysis and improved comprehension. The authors propose a model-driven approach that uses DMN models to generate Machine Learning (ML) training data and show, how the trained ML models can inform human decision modelers by means of superimposing the feature importance within the original DMN models. An evaluation with multiple real DMN models from an insurance company evaluates the feasibility and the utility of the approach.
  • Zeitschriftenartikel
    An ETA Prediction Model for Intermodal Transport Networks Based on Machine Learning
    (Business & Information Systems Engineering: Vol. 62, No. 5, 2020) Balster, Andreas; Hansen, Ole; Friedrich, Hanno; Ludwig, André
    Transparency in transport processes is becoming increasingly important for transport companies to improve internal processes and to be able to compete for customers. One important element to increase transparency is reliable, up-to-date and accurate arrival time prediction, commonly referred to as estimated time of arrival (ETA). ETAs are not easy to determine, especially for intermodal freight transports, in which freight is transported in an intermodal container, using multiple modes of transportation. This computational study describes the structure of an ETA prediction model for intermodal freight transport networks (IFTN), in which schedule-based and non-schedule-based transports are combined, based on machine learning (ML). For each leg of the intermodal freight transport, an individual ML prediction model is developed and trained using the corresponding historical transport data and external data. The research presented in this study shows that the ML approach produces reliable ETA predictions for intermodal freight transport. These predictions comprise processing times at logistics nodes such as inland terminals and transport times on road and rail. Consequently, the outcome of this research allows decision makers to proactively communicate disruption effects to actors along the intermodal transportation chain. These actors can then initiate measures to counteract potential critical delays at subsequent stages of transport. This approach leads to increased process efficiency for all actors in the realization of complex transport operations and thus has a positive effect on the resilience and profitability of IFTNs.
  • Zeitschriftenartikel
    Auswirkungen der Medizinprodukteverordnung auf ML-Lösungen in Schweizer Spitälern
    (HMD Praxis der Wirtschaftsinformatik: Vol. 61, No. 2, 2024) Russ, Christian; Stalder, Philipp H.; Rufinatscha, Stefanie; Pimentel, Tibor; Geissmann, Lukas
    Künstliche Intelligenz (KI) ist schon länger in den Spitälern direkt und indirekt präsent. Oftmals ist KI für Arbeitsplatzfunktionen im Bürobereich wie z. B. in Spracherkennungssoftware verfügbar, teilweise auch in Personal- und Ressourcen-Optimierungssoftware. Das Spektrum reicht speziell im medizinischen Bereich von datengetriebenen Analysen und Informationsunterstützungssystemen bis hin zur Generierung von Diagnose- und Therapievorschlägen für das medizinische Personal. Jedoch sind vielen Akteuren in den Spitälern der Umfang und die Auswirkung von KI-Technologien gar nicht wirklich bewusst. Noch weniger bekannt sind dabei die regulatorischen Vorgaben in Kombination mit dem Einsatz von Maschinellem Lernen (ML). Basierend auf einer repräsentativen Befragung von allgemeinen Spitälern in der Schweiz wurde der aktuelle Stand der KI-Nutzung erhoben. Auf dieser Basis werden die Anforderungen an ML-Systeme in Bezug auf die Medizinprodukteverordnung und deren Auswirkung in Hinblick auf den konformen Einsatz von medizinischer Software analysiert. Wir präsentieren einen Vorschlag, wie ML-Systeme besser mit den Regulatorien in Einklang gebracht werden können. Im Ausblick wird auf die möglichen Grenzen und Notwendigkeiten für zukünftige Weiterentwicklungen eingegangen. Artificial intelligence (AI) has been present in hospitals directly and indirectly for some time. Often AI is available for workplace functions in the office area, such as speech recognition software, and in some cases also in personnel and resource optimization software. In the medical field, specifically, the spectrum ranges from data-driven analyses and information support systems to the generation of diagnostic and therapeutic suggestions for medical personnel. However, many players in hospitals are not aware of the scope and impact of AI technologies. What is not well known are the regulatory requirements in combination with the use of machine learning (ML). Based on a representative survey of general hospitals in Switzerland, the current state of AI usage was determined. On this basis, we analyze the requirements for ML systems with respect to the Medical Device Regulation and their impact with respect to the compliant use of medical software. We present a proposal on how ML systems can be brought more in line with regulations. In the concluding outlook, the possible limitations and necessities for future developments are discussed.
  • Zeitschriftenartikel
    Automatic Detection of Visual Search for the Elderly using Eye and Head Tracking Data
    (KI - Künstliche Intelligenz: Vol. 31, No. 4, 2017) Dietz, Michael; Schork, Daniel; Damian, Ionut; Steinert, Anika; Haesner, Marten; André, Elisabeth
    With increasing age we often find ourselves in situations where we search for certain items, such as keys or wallets, but cannot remember where we left them before. Since finding these objects usually results in a lengthy and frustrating process, we propose an approach for the automatic detection of visual search for older adults to identify the point in time when the users need assistance. In order to collect the necessary sensor data for the recognition of visual search, we develop a completely mobile eye and head tracking device specifically tailored to the requirements of older adults. Using this device, we conduct a user study with 30 participants aged between 65 and 80 years ($$avg = 71.7,$$avg=71.7, 50% female) to collect training and test data. During the study, each participant is asked to perform several activities including the visual search for objects in a real-world setting. We use the recorded data to train a support vector machine (SVM) classifier and achieve a recognition rate of 97.55% with the leave-one-user-out evaluation method. The results indicate the feasibility of an approach towards the automatic detection of visual search in the wild.
  • Zeitschriftenartikel
    Automatisierung von Geschäftsprozessen im Maschinen- und Anlagenbau – Fallstudie zu Predictive Maintenance
    (HMD Praxis der Wirtschaftsinformatik: Vol. 56, No. 5, 2019) Gluchowski, Peter; Schieder, Christian; Gmeiner, Andreas; Trenz, Stefan
    Die Chancen, die sich durch die zielgerichtete Auswertung und Verwendung von Sensordaten für den Maschinen- und Anlagenbau ergeben, sind immens. Große Anlagen weisen hunderte oder gar tausende von verbauten Sensoren auf, die in kurzen Zeitabständen Daten über aktuelle Zustände einzelner Maschinenkomponenten sowie der Produktionsprozesse erzeugen. Die Produktion von Wellpappe, die als vielseitiges Verpackungsmaterial für Endkunden- und Industrieprodukte weltweit zum Einsatz kommt, stellt hierbei ein besonders anschauliches Beispiel dar. Die Entwicklung digitaler Dienstleistungen wie die vorausschauende Wartung (sog. „Predictive Maintenance“) basieren auf Daten, die an der Anlage erzeugt werden. Ein im Produktionsprozess von Wellpappe kritisches Bauteil stellt das bei der Verklebung der Wellpapp-Schichten verwendete Anpressband dar. Die neueste Generation von Wellpappenanlagen wird zu diesem Zweck mit spezieller Sensorik ausgestattet, die laufend Daten zum Zustand des Bandes liefern. Mit diesen Daten lassen sich mit Hilfe modellbasierter maschineller Lernverfahren Prognosen zur Lebensdauer treffen und damit Automatisierungspotenziale bei nachfolgenden Geschäftsprozessen ausschöpfen. Ziel ist die Minimierung der Produktions- und Qualitätsverluste sowie die Automatisierung der Ersatzteilprozesse. Der Beitrag skizziert die Vorgehensweise und Ergebnisse des zugehörigen Projekts und gibt einen Ausblick auf zukünftige Entwicklungen. The systematic evaluation and use of sensor data create immense opportunities for mechanical engineering and machine operations. Production lines have hundreds or even thousands of built-in sensors that generate data on the current status of individual machine components and production processes at short intervals. The production of corrugated board, which is used worldwide as a versatile packaging material for end customer and industrial products, is a particularly vivid example of this. The development of digital services such as predictive maintenance is based on data generated at the production line. A critical component in the production process of corrugated board is the pressure belt used to bond the layers of corrugated board. For this purpose, the latest generation of corrugators is equipped with special sensors that continuously provide data on the condition of the belt. These data can be used to predict the service life with the aid of model-based machine learning methods and thus exploit automation potential in subsequent business processes. The aim is to minimize production and quality losses and automate spare parts processes. The paper outlines the approach and results of the associated project and gives an outlook on future developments.
  • Zeitschriftenartikel
    Beyond Reinforcement Learning and Local View in Multiagent Systems
    (KI - Künstliche Intelligenz: Vol. 28, No. 3, 2014) Bazzan, Ana L. C.
    Learning is an important component of an agent’s decision making process. Despite many messages in contrary, the fact is that, currently, in the multiagent community it is mostly likely that learning means reinforcement learning. Given this background, this paper has two aims: to revisit the “old days” motivations for multiagent learning, and to describe some of the work addressing the frontiers of multiagent systems and machine learning. The intention of the latter task is to try to motivate people to address the issues that are involved in the application of techniques from multiagent systems in machine learning and vice-versa.
  • Zeitschriftenartikel
    Connecting Question Answering and Conversational Agents
    (KI - Künstliche Intelligenz: Vol. 26, No. 4, 2012) Waltinger, Ulli; Breuing, Alexa; Wachsmuth, Ipke
    Research results in the field of Question Answering (QA) have shown that the classification of natural language questions significantly contributes to the accuracy of the generated answers. In this paper we present an approach which extends the prevalent question classification techniques by additionally considering further contextual information provided by the questions. Thereby we focus on improving the conversational abilities of existing interactive interfaces by enhancing their underlying QA systems in terms of response time and correctness. As a result, we are able to introduce a method based on a tripartite contextualization. First, we present a comprehensive question classification experiment based on machine learning using two different datasets and various feature sets for the German language. Second, we propose a method for detecting the focus chunk of a given question, that is, for identifying which part of the question is fundamentally relevant to the answer and which part refers to a specification of it. Third, we investigate how to identify and label the topic of a given question by means of a human-judgment experiment. We show that the resulting contextualization method contributes to an improvement of existing question answering systems and enhances their application within interactive scenarios.
  • Konferenzbeitrag
    Counterfactual Explanations for Models of Code
    (Software Engineering 2024 (SE 2024), 2024) Cito, Jürgen; Dillig, Isil; Murali, Vijayaraghavan; Chandra, Satish