Our workshop aims to bring together researchers and practitioners from the fields of AI and/or production investigating, developing, or exploring AI techniques in production. We aim to provide a platform for the exchange of ideas and experiences under the general topic of ‘AI in Production’, not specializing in certain fields of production nor AI but explicitly including production planning, control and optimization. Ideally, our workshop will enable us to standardize approaches for supporting production applying AI or to transfer these approaches from one area of application in production to another. Thus, the Workshop is not only intended for experts in artificial intelligence (in production), but explicitly also for professionals from production.:This Proceedings on 'AI in Production' consists of 5 Proceedings: - Charging Strategies for Automated Guided Vehicles Using Supervised Learning - Optical Neural Networks for Low-latency and Energy-efficient Applications in Production - Flexible Data Architecture for Enabling AI Applications in Production Environments - Perception of Biases in Machine Learning in Production Research - A Structured Literature Review Dissecting Bias Categories - Supporting machine operators in paper production using machine learning based state estimation and user assistance system

Constructing a Bayesian Network requires the conditional probabilities table (CPT) to be acquired, one for each variable or node in the network. When data mining is not available, CPTs must be acquired from the domain experts. The complexity of the direct elicitation is exponential on the number of parents of a variable, making direct elicitation from human experts impractical for a large number of causes. Causal models such as Noisy-OR, Noisy-AND, Noisy-MIN, Noisy-MAX and Recursive Noisy-OR have been developed that allow CPTs acquisition to be achieved with linear complexity on the number of causes. Their representation power is measured by their ability to encode the causal interactions. Causal interactions can be categorized into two types: reinforcing and undermining. The Non-Impeding Noisy-AND or NIN-AND tree causal model, developed by Xiang and Jia, is capable of modeling both types of interaction while retaining the linear complexity. The main challenge in utilizing the NIN-AND tree model to generate a CPT is that it requires its tree topology to be elicited. A NIN-AND tree topology is an encoding of the causal interactions between the causes. In this work we present two methods, Structure Elimination (SE) and Pairwise Causal Interaction (PCI), that allow indirect elicitations of the NIN-AND tree topology using some additional probabilities elicited from experts. We conduct human-based experiment to investigate the e ectiveness of the two methods in terms of accuracy by comparing them to the Direct Numerical (DN) elicitation method. We recruit participants from second year Computer Science students at the University of Guelph. The process involves training a participant into domain expert using a known NIN-AND tree model then acquire another NIN-AND tree model by applying the SE and PCI methods. The CPTs produced by the acquired NIN-AND tree models are then compared to the one obtained by using the DN method. Comparable CPT accuracies are obtained among models generated by di erent methods, even though SE and PCI requires a much smaller number of parameters in comparison to DN.

The Internet of Things (IoT) has deeply changed how we interact with our world. Today, smart homes, self-driving cars, connected industries, and wearables are just a few mainstream applications where IoT plays the role of enabling technology. When IoT became popular, Cloud Computing was already a mature technology able to deliver the computing resources necessary to execute heavy tasks (e.g., data analytic, storage, AI tasks, etc.) on data coming from IoT devices, thus practitioners started to design and implement their applications exploiting this approach. However, after a hype that lasted for a few years, cloud-centric approaches have started showing some of their main limitations when dealing with the connectivity of many devices with remote endpoints, like high latency, bandwidth usage, big data volumes, reliability, privacy, and so on. At the same time, a few new distributed computing paradigms emerged and gained attention. Among all, Edge Computing allows to shift the execution of applications at the edge of the network (a partition of the network physically close to data-sources) and provides improvement over the Cloud Computing paradigm. Its success has been fostered by new powerful embedded computing devices able to satisfy the everyday-increasing computing requirements of many IoT applications. Given this context, how can next-generation IoT applications take advantage of the opportunity offered by Edge Computing to shift the processing from the cloud toward the data sources and exploit everyday-more-powerful devices? This thesis provides the ingredients and the guidelines for practitioners to foster the migration from cloud-centric to novel distributed design approaches for IoT applications at the edge of the network, addressing the issues of the original approach. This requires the design of the processing pipeline of applications by considering the system requirements and constraints imposed by embedded devices. To make this process smoother, the transition is split into different steps starting with the off-loading of the processing (including the Artificial Intelligence algorithms) at the edge of the network, then the distribution of computation across multiple edge devices and even closer to data-sources based on system constraints, and, finally, the optimization of the processing pipeline and AI models to efficiently run on target IoT edge devices. Each step has been validated by delivering a real-world IoT application that fully exploits the novel approach. This paradigm shift leads the way toward the design of Edge Intelligence IoT applications that efficiently and reliably execute Artificial Intelligence models at the edge of the network.

No description available.

No description available.

AbstraktSyfte: Syftet med studien var att analysera hur företag kan förbättra förståelsen för denekonomiska lönsamheten i sina marknadsföringsaktiviteter, detta genom att undersöka hurföretag idag arbetar med att följa upp den lönsamheten av digital marknadsföring samtvilken förbättringspotential företag upplever att det finns kring sätten de arbetar på. Metod: En kvalitativ metod har genomförts i denna studie där fyra intervjuer hargenomförts med tre respondenter som arbetar med att utvärdera lönsamhet av digitalmarknadsföring. Den insamlade empirin har sedan jämförts med nuvarande teorier för attförsöka förbättra förståelsen för ämnet. Resultat: Företag använder idag främst ROAS för att mäta lönsamheten avmarknadsföring och att den data som utgör grunden för utvärderingen kommer ochbearbetas av Google och Metas AI modeller. Däremot behandlas dessa olika av företag,vilket har visat sig motsvara deras erfarenhet och expertis inom området. De mest väsentliga aspekter som uppmärksammades var förbättringen och utnyttjandet avAI och dataanalysverktyg då den är kärnan i att utvärdera lönsamheten avmarknadsföringsaktiviteterna. Respondenterna är övertygade om att AI kommer utgöra enallt större del av arbetet med att utvärdera lönsamheten med digital marknadsföring därförmågan att förstå och hantera dessa AI modeller kommer att vara viktigt för företag iframtiden.Slutsats: Studien beskriver hur företag arbetar i praktiken och bidrar till nuvarande teoriergenom att visa på områden som bekräftar och säger emot tidigare forskning. Till exempelhar studien bekräftat problemet med att kunden inte delar med sig av information på nätet,vilket styrks av litteraturen. Medan det också visat sig att företagen i studien endast harutvärderat lönsamhetsmåttet ROAS, vilket säger emot tidigare forskning. / Purpose: The purpose of the study was to analyze how companies can improve theirunderstanding of the economic profitability of their marketing activities by examining howcompanies currently work to follow up on the profitability of digital marketing, as well asthe potential for improvement that companies perceive in the ways they work. Method: A qualitative method has been employed in this study, where four interviewshave been conducted with three respondents who work on evaluating the profitability ofdigital marketing. The collected empirical data has then been compared with currenttheories in order to enhance understanding of the subject. Results: Companies today primarily use ROAS (Return on Advertising Spend) to measurethe profitability of marketing, and the data that forms the basis for evaluation is providedand processed by Google and Meta's AI models. However, these are treated differently bycompanies, which has been found to correspond to their experience and expertise in thefield. The most significant aspects that were highlighted were the improvement and utilization ofAI and data analysis tools, as they are at the core of evaluating the profitability ofmarketing activities. The respondents are convinced that AI will play an increasinglyimportant role in evaluating the profitability of digital marketing, where the ability tounderstand and manage these AI models will be crucial for companies in the future.

No description available.

No description available.

Inwieweit KI das neuartige universitäre Lehren und Lernen wirksam begleiten kann, wird im BMBF-Verbundprojekt 'tech4comp: Personalisierte Kompetenzentwicklung durch skalierbare Mentoringprozesse' untersucht. Gemeinsam beforscht man soziotechnische Artefakte für personalisiertes digital-gestütztes Mentoring für Studierende. Hierzu werden u.a. Rahmenbedingungen und (soziale) Kontextfaktoren erforscht, um die Implementierung von KI in der Hochschulbildung zu unterstützen. Es wird davon ausgegangen, dass unabhängig von der Art der Technologie und vom pandemischen Kontext, insbesondere die Akzeptanz und Bereitschaft der beteiligten Stakeholder zum erfolgreichen Einsatz intelligenter Bildungstechnologien beiträgt. Das ZQA/KfBH der TU Dresden widmet sich unter der Leitung von Dr. Cathleen M. Stützer im Forschungsprojekt der Elaboration von Handlungsfeldern, die sich aus einer soziotechnischen Beforschung von KI in der Hochschulbildung ergeben. Fallstudien hierzu stellen sich u. a. Fragen zu Gelingensbedingungen und Wirksamkeit digitaler Hochschulbildung, um (prospektiv) eine erfolgreiche Implementierung KI-gestützter adaptiver Mentoringsysteme mit evidenten Forschungsberichten zu unterstützen.:Vorwort & Danksagung Abbildungsverzeichnis Tabellenverzeichnis Abkürzungsverzeichnis 1. Einleitung 2. Methodik 3. Ergebnisse 4. Implikationen 4.1 Einflussfaktoren und Gelingensbedingungen der KI-Akzeptanz 4.2 Handlungsempfehlungen 5. Zusammenfassung und Fazit 6. Limitationen 7. Literaturverzeichnis Anhang / The extent to which AI can effectively accompany new types of university teaching and learning is being investigated in the BMBF joint project 'tech4comp: Personalised competence development through scalable mentoring processes'. Together, they are researching socio-technical artefacts for personalised digitally-supported mentoring for students. For this purpose, framework conditions and (social) contextual factors, among others, are being researched in order to support the implementation of AI in higher education. It is assumed that regardless of the type of technology and the pandemic context, the acceptance and willingness of the stakeholders involved in particular contributes to the successful use of intelligent educational technologies. Under the direction of Dr. Cathleen M. Stützer, the ZQA/KfBH at TU Dresden is dedicated to the elaboration of fields of action resulting from socio-technical research on AI in higher education. Case studies on this topic address questions such as the conditions for success and the effectiveness of digital higher education in order to (prospectively) support the successful implementation of AI-supported adaptive mentoring systems with evident research reports.:Vorwort & Danksagung Abbildungsverzeichnis Tabellenverzeichnis Abkürzungsverzeichnis 1. Einleitung 2. Methodik 3. Ergebnisse 4. Implikationen 4.1 Einflussfaktoren und Gelingensbedingungen der KI-Akzeptanz 4.2 Handlungsempfehlungen 5. Zusammenfassung und Fazit 6. Limitationen 7. Literaturverzeichnis Anhang

Syftet med denna studie är att undersöka hur AI gestaltas i tidningarna Resumé och Dagens Media som är inriktade på media- och kommunikation. Detta uppnås genom att studera tidningarnas artiklar om AI. Studien undersöker vilka gestaltningar som förekommer i artiklarna och om tidningarna uppfyller UNESCO:s riktlinjer om hur journalister bör rapportera om AI. De gestaltningar som undersöks är sociala framsteg, Pandora´s box, ekonomisk utveckling/konkurrenskraft, utfallsframe och tidsperspektiv. Det undersöks även hur ofta AI ersätter syssla förekommer. De riktlinjer från UNESCO som undersöktes var AI:s begränsningar, det mänskliga arbetet, land, andra röster, suggestiva bilder och miljöaspekter. Det teoretiska ramverk som studien bygger på är gestaltningsteorin och teknologisk determinism. Studiens resultat visar att Resumé och Dagens Media Resumé tenderar att skildra AI som eftersträvansvärt genom att beskriva konsekvenserna av att använda AI som något läsaren vinner på. AI gestaltas även som en teknik som, helt eller delvis, kommer ersätta sysslor och arbetsuppgifter i mediabranschen. De variabler som kommer från UNESCO:s riktlinjer förekommer generellt sätt mindre. Resultatet visar att tidningarna skildrar AI som en teknik som utvecklas av sig själv, inte går att kontrollera, har få begränsningar, inte får mycket kritik, inte påverkar miljön och kan likställas med en människa. Således anammar de teorin teknologisk determinism, och i synnerhet de typer som kallas justificatory och normative. / The purpose of this study is to examine how AI is framed in Swedish news media oriented toward media and communication. This is achieved by studying the magazines Resumé and Dagens Media´s articles about AI. This study examines which frames can be found in the articles and if the magazines manage to fulfill recommended guidelines created by UNESCO on how journalists should report on AI. The frames that are examined are social progress, Pandora´s box, economic development/competitiveness, outcome frame and temporal frame. It is also examined how often AI replaces a task.  The UNESCO guidelines that are examined are the limitations of AI, the human labor behind AI, country, other voices, suggestive images and environmental aspects. The theoretical frameworks the study is built upon are framing theory and technological determinism. The study´s results show that Resumé and Dagens Media tend to portray AI as something desirable by describing the consequences of using AI as something the reader will benefit from. AI is also portrayed as a technology that will, completely or partially, replace tasks within the media industry. The variables that derive from the UNESCO guidelines generally occur less. The result shows that the magazines portray AI as a technology that can evolve by itself, cannot be controlled, have few limitations, does not get much critique, does not affect the environment, and can be equated with a human being. Therefore, the theory technological determinism is applied by the magazines, especially the types that are called justificatory and normative.