Exact Combinatorial Optimization with Graph Convolutional Neural Networks

Combinatorial optimization problems are typically tackled by the branch-and-bound paradigm. We propose to learn a variable selection policy for branch-and-bound in mixed-integer linear programming, by imitation learning on a diversified variant of the strong branching expert rule. We encode states as bipartite graphs and parameterize the policy as a graph convolutional neural network. Experiments on a series of synthetic problems demonstrate that our approach produces policies that can improve upon expert-designed branching rules on large problems, and generalize to instances significantly larger than seen during training.

Development and scale-up studies of bio-based poly(ester-amide)s

Plastics are polymers of conventional and extensive use in our day-to-day life. This is due to their light weight, adaptability to different uses and low prices. A downside of such extensive use is the environmental pollution arising from plastic production and disposal. Indeed, many commodity polymers are produced from non-renewable resources while other do not bio-degrade after their end-of-life disposal. Consequently, the ideal polymer comes from renewable raw materials and bio-degrades after its disposal, meaning that it would do little or no harm to the environment from the beginning to the end of its life cycle. In this thesis project a class of bio-based and bio-degradable co-polymers, namely poly(ester-amide)s, was investigated because of their tunable mechanical and bio-degradation properties as well as their renewable origin. Such polymers were synthetized and characterized thermically and mechanically. Furthermore, a scale-up procedure was developed and applied to one polymer and processing trials were made with the material obtained after scale-up.