Abstract Software-Defined Networking (SDN) is an emerging network architecture characterized by the decoupling of the data plane from the control plane, as well as managing the whole network in a centralized logic. As the scale of networks continues to expand, the single-controller architecture is no longer able to meet the performance and reliability requirements of the network. Consequently, a logically centralized while physically distributed multi-controller architecture has been proposed, in which the number and locations of controllers must be determined, formulating the Controller Placement Problem (CPP). In order to solve the CPP and optimize the propagation latency, we propose a Convolutional Node Embedding and K-means Algorithm (CNEKA), which integrates Graph Convolutional Networks (GCN) with the K-means algorithm. More specifically, we adopt the principles of GCN to facilitate mutual information propagation among adjacent nodes, and then use the K-means algorithm to achieve graph segmentation through the low-dimensional embedding vectors calculated by GCN. The study demonstrates that the CNEKA algorithm significantly enhances performance in optimizing average and worst-case latency between controllers and switches, as well as overall network latency, particularly under high controller counts. The algorithm achieves a mere 1.10% deviation from the global optimum in average controller-switch latency, underscoring its high efficacy.
Although part-based motion synthesis networks have been investigated to reduce the complexity of modeling heterogeneous human motions, their computational cost remains prohibitive in interactive applications. To this end, we propose a novel two-part transformer network that aims to achieve high-quality, controllable motion synthesis results in real-time. Our network separates the skeleton into the upper and lower body parts, reducing the expensive cross-part fusion operations, and models the motions of each part separately through two streams of auto-regressive modules formed by multi-head attention layers. However, such a design might not sufficiently capture the correlations between the parts. We thus intentionally let the two parts share the features of the root joint and design a consistency loss to penalize the difference in the estimated root features and motions by these two auto-regressive modules, significantly improving the quality of synthesized motions. After training on our motion dataset, our network can synthesize a wide range of heterogeneous motions, like cartwheels and twists. Experimental and user study results demonstrate that our network is superior to state-of-the-art human motion synthesis networks in the quality of generated motions.
Although part-based motion synthesis networks have been investigated to reduce the complexity of modeling heterogeneous human motions, their computational cost remains prohibitive in interactive applications. To this end, we propose a novel two-part transformer network that aims to achieve high-quality, controllable motion synthesis results in real-time. Our network separates the skeleton into the upper and lower body parts, reducing the expensive cross-part fusion operations, and models the motions of each part separately through two streams of auto-regressive modules formed by multi-head attention layers. However, such a design might not sufficiently capture the correlations between the parts. We thus intentionally let the two parts share the features of the root joint and design a consistency loss to penalize the difference in the estimated root features and motions by these two auto-regressive modules, significantly improving the quality of synthesized motions. After training on our motion dataset, our network can synthesize a wide range of heterogeneous motions, like cartwheels and twists. Experimental and user study results demonstrate that our network is superior to state-of-the-art human motion synthesis networks in the quality of generated motions.
Mixing limestone powder (LP) in the self-leveling mortar (SLM) can not only solve the problems of LP waste randomly piled up and secondary utilization of resources, but also reduce the raw material cost of SLM and have excellent mechanical properties. The effect of replacing fly ash (FA) with LP and replacing cement with LP after completely replacing FA on fluidity and strength of SLM are studied. The microstructure of SLM is measured by mercury intrusion porosimetry and scanning electron microscope. The results show that the initial fluidity and the 20-min fluidity of SLM decrease gradually with the increase of LP content. The strength of SLM increases and then decreases with the increase of LP replacing FA, and the strength is the highest when the addition of LP is 40%. When LP replaces cement after completely replacing FA, the strength of SLM decreases with the increase of displacement. Excessive LP can greatly damage the mechanical properties of SLM. The appropriate content of LP can improve the microstructure of SLM and promote the formation of hydration products, which is helpful to reduce the porosity and thus improves the structure density. This may be due to the chemical reaction and the microfiller effect of LP.
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)