Kongsberg Defence & Aerospace – a Norwegian multinational company that develops and delivers innovative technologies to safeguard people and critical infrastructure – is experiencing a fast-growing order backlog and a growing need for production managers to ensure efficient, high-quality and competitive production processes in its factories around the world. Recognizing this need, and through the establishment of a professorship in production management at the University of South-Eastern Norway, the company is seeking to establish a world class education program in production and operations management to develop and strengthen education in the field. As such, this paper sets out to benchmark production and operations management programs across the member institutions of the IFIP working group 5.7: Advances in Production Management Systems. We analyze both bachelor and master programs in the fields of engineering and management from 16 institutions spanning twelve different countries: Norway, the Netherlands, Mexico, Italy, Brazil, France, Denmark, Sweden, Switzerland, Australia, the UK and the USA. We also present the findings of an industrial focus group initiative in which participants considered the results of the global benchmarking study and identified areas for development based on gaps in the current offerings. The findings can be used by member institutions to help navigate current and future geopolitical challenges.

When designing Convolutional Neural Networks (CNNs), one must select the size of the convolutional kernels before training. Recent works show CNNs benefit from different kernel sizes at different layers, but exploring all possible combinations is unfeasible in practice. A more efficient approach is to learn the kernel size during training. However, existing works that learn the kernel size have a limited bandwidth. These approaches scale kernels by dilation, and thus the detail they can describe is limited. In this work, we propose FlexConv, a novel convolutional operation with which high bandwidth convolutional kernels of learnable kernel size can be learned at a fixed parameter cost. FlexNets model long-term dependencies without the use of pooling, achieve state-of-the-art performance on several sequential datasets, outperform recent works with learned kernel sizes, and are competitive with much deeper ResNets on image benchmark datasets. Additionally, FlexNets can be deployed at higher resolutions than those seen during training. To avoid aliasing, we propose a novel kernel parameterization with which the frequency of the kernels can be analytically controlled. Our novel kernel parameterization shows higher descriptive power and faster convergence speed than existing parameterizations. This leads to important improvements in classification accuracy.