Mercury-ML
Mercury-ML is an open source Machine Learning workflow management library. Its core contributors are employees of Alexander Thamm GmbH …

Differentiable Greedy Network (DGN)
Optimal selection of a subset of items from a given set is a hard problem that requires combinatorial optimization. In this paper, we propose a subset selection algorithm that is trainable with gradient-based methods yet achieves near-optimal performance via submodular optimization. We focus on the task of identifying a relevant set of sentences for claim verification in the context of the FEVER task. Conventional methods for this task look at sentences on their individual merit and thus do not optimize the informativeness of sentences as a set. We show that our proposed method which builds on the idea of unfolding a greedy algorithm into a computational graph allows both interpretability and gradient-based training. The proposed differentiable greedy network (DGN) outperforms discrete optimization algorithms as well as other baseline methods in terms of precision and recall. …

Balanced k-Means
Mesh partitioning is an indispensable tool for efficient parallel numerical simulations. Its goal is to minimize communication between the processes of a simulation while achieving load balance. Established graph-based partitioning tools yield a high solution quality; however, their scalability is limited. Geometric approaches usually scale better, but their solution quality may be unsatisfactory for `non-trivial’ mesh topologies. In this paper, we present a scalable version of $k$-means that is adapted to yield balanced clusters. Balanced $k$-means constitutes the core of our new partitioning algorithm Geographer. Bootstrapping of initial centers is performed with space-filling curves, leading to fast convergence of the subsequent balanced k-means algorithm. Our experiments with up to 16384 MPI processes on numerous benchmark meshes show the following: (i) Geographer produces partitions with a lower communication volume than state-of-the-art geometric partitioners from the Zoltan package; (ii) Geographer scales well on large inputs; (iii) a Delaunay mesh with a few billion vertices and edges can be partitioned in a few seconds. …

BlockPuzzle
In this work we propose a novel task framework under which a variety of physical reasoning puzzles can be constructed using very simple rules. Under sparse reward settings, most of these tasks can be very challenging for a reinforcement learning agent to learn. We build several simple environments with this task framework in Mujoco and OpenAI gym and attempt to solve them. We are able to solve the environments by designing curricula to guide the agent in learning and using imitation learning methods to transfer knowledge from a simpler environment. This is only a first step for the task framework, and further research on how to solve the harder tasks and transfer knowledge between tasks is needed. …