Both Software Engineering and Machine Learning have become recognized disciplines. In this article I analyse the combination of the two: engineering of machine learning applications. I believe the systematic way of working for machine learning applications is at certain points different from traditional (rule-based) software engineering. The question I set out to investigate is “How does software engineering change when we develop machine learning applications”?. This question is not an easy to answer and turns out to be a rather new, with few publications. This article collects what I have found until now.
LINK
Data-driven condition-based maintenance (CBM) and predictive maintenance (PdM) strategies have emerged over recent years and aim at minimizing the aviation maintenance costs and environmental impact by the diagnosis and prognosis of aircraft systems. As the use of data and relevant algorithms is essential to AI-based gas turbine diagnostics, there are different technical, operational, and regulatory challenges that need to be tackled in order for the aeronautical industry to be able to exploit their full potential. In this work, the machine learning (ML) method of the generalised additive model (GAM) is used in order to predict the evolution of an aero engine’s exhaust gas temperature (EGT). Three different continuous synthetic data sets developed by NASA are employed, known as New Commercial Modular Aero-Propulsion System Simulation (N-CMAPSS), with increasing complexity in engine deterioration. The results show that the GAM can be predict the evolution of the EGT with high accuracy when using several input features that resemble the types of physical sensors installed in aero gas turbines currently in operation. As the GAM offers good interpretability, this case study is used to discuss the different data attributes a data set needs to have in order to build trust and move towards certifiable models in the future.
This paper presents a Decision Support System (DSS) that helps companies with corporate reputation (CR) estimates of their respective brands by collecting provided feedbacks on their products and services and deriving state-of-the-art key performance indicators. A Sentiment Analysis Engine (SAE) is at the core of the proposed DSS that enables to monitor, estimate, and classify clients’ sentiments in terms of polarity, as expressed in public comments on social media (SM) company channels. The SAE is built on machine learning (ML) text classification models that are cross-source trained and validated with real data streams from a platform like Trustpilot that specializes in user reviews and tested on unseen comments gathered from a collection of public company pages and channels on a social networking platform like Facebook. Such crosssource opinion analysis remains a challenge and is highly relevant in the disciplines of research and engineering in which a sentiment classifier for an unlabeled destination domain is assisted by a tagged source task (Singh and Jaiswal, 2022). The best performance in terms of F1 score was obtained with a multinomial naive Bayes model: 0,87 for validation and 0,74 for testing.
Granular materials (GMs) are simply a collection of individual particles, e.g., rice, coffee, iron-ore. Although straightforward in appearance, GMs are key to several processes in chemical-pharmaceutical, high-tech, agri-food and energy industry. Examples include laser sintering in additive manufacturing, tableting in pharma or just mixing of your favourite crunchy muesli mix in food industry. However, these bulk material handling processes are notorious for their inefficiency and ineffectiveness. Thereby, affecting the overall expenses and product quality. To understand and enhance the quality of a process, GMs industries utilise computer-simulations, much like how cars and aeroplanes have been designed and optimised since the 1990s. Just as how engineers utilise advanced computer-models to develop our fuel-efficient vehicle design, energy-saving granular processes are also developed utilising physics-based simulation-models, using a computer. Although physics-based models can effectively optimise large-scale processes, creating and simulating a fully representative virtual prototype of a GMs process is very iterative, computationally expensive and time intensive. On the contrary, given the available data, this is where machine learning (ML) could be of immense value. Like how ML has transformed the healthcare, energy and other top sectors, recent ML-based developments for GMs show serious promise in faster virtual prototyping and reduced computational cost. Enabling industries to rapidly design and optimise, enhancing real-time data-driven decision making. GranML aims to empower the GMs industries with ML. We will do so by (i) performing an in-depth GMs-ML literature review, (ii) developing open-access ML implementation guidelines; and (iii) an open-source proof-of-concept for an industry-relevant use case. Eventually, our follow-up mission is to build upon this vital knowledge by (i) expanding the consortium; (ii) co-developing a unified methodology for efficient computer-prototyping, unifying physics- and ML-based technologies for GMs; (iii) enhancing the existing computer-modelling infrastructure; and (iv) validating through industry focused demonstrators.
Volgens onderzoek van McKinsey is marketing het vakgebied waar AI de meeste waarde toe gaat voegen, onder andere op het gebied van personalisatie. Hierdoor verandert het stakeholderveld waarin de marketeer personalisatie-algoritmen toepast significant, zo werkt hij steeds vaker samen met data scientists, AI-architecten en data engineers. Dit onderzoek richt zich op de vraag welke handvatten marketeers nodig hebben om tot een verantwoorde personalisatie-praktijk te komen.
With increasing labor shortages, sectors using mobile machines (automotive/industry/agrifood/logistics) have a rising need for productivity improvement. With evolving technology, mobile machine control has stepped from hydraulics to electronics using sensors and smart systems to support drivers and allowing intelligent and automated machine functions. Verification and validation costs of such complex functionality urge the need for virtual solution routes to limit the lead time, cost and safety issues of real-world testing. RAAK-mkb project Fast&Curious developed tools to enable model-driven development for the control of a wide range of vehicle systems. This included automatic code generation support from MATLAB/Simulink® into the Bodas RC30 family vehicle controllers from Bosch Rexroth (see www.openMBD.com). The solution has been adopted by several SMEs allowing them to start working in a model-driven way, helping them to do virtual verification&validation, lowering development time and costs. Meanwhile, Rexroth adopted MATLAB/Simulink for core vehicle functions development and currently develops Fast&Curious-alike automatic code generation support for their recent RC40 controllers. Virtuoso aims to further improve productivity on simulation level by creating an interface layer in Simulink to (automatically) test impact of hardware interface imperfections and failures, such as noise and short circuits, as well as to seamlessly switch between continuous (early development) and discretized (deployment-oriented) input/output behavior. Companies like Emoss and Jautomatisering are interested in such solutions, allowing them to adopt efficient, model-driven processes and supporting their engineers in the required hydraulics-to-software/electronics skill-shift. The solution connects well to future developments like robotization. Besides supporting development of vehicle automation and mobile robotics, MATLAB/Simulink also supports ROS (Robot Operating System) via co-simulation and co-deployment. ROS has become the standard in (mobile) robot control development and is used by many parties. Virtuoso further closes the gap between development and deployment and allows future integration in mobile robotics, foreseen as next step.
