As the need for on-device machine learning is increasing recently, embedded devices tend to be equipped with heterogeneous processors that include a multi-core CPU, a GPU, and/or a DNN accelerator called a Neural Processing Unit (NPU). In the scheduling of multiple deep learning (DL) applications in such embedded devices, there are several technical challenges. First, a task can be mapped onto a single core or any number of available cores. So we need to consider various possible configurations of CPU cores. Second, embedded devices usually apply Dynamic Voltage and Frequency Scaling (DVFS) to reduce energy consumption at run-time. We need to consider the effect of DVFS in the profiling of task execution times. Third, to avoid overheat condition, it is recommended to limit the core utilization. Lastly, some cores will be shut-down at run-time if core utilization is not high enough, in case the hot-plugging option is turned on. In this paper, we propose a scheduling technique based on Genetic Algorithm to run DL applications on heterogeneous processors, considering all those issues. First, we aim to optimize the throughput of a single deep learning application. Next, we aim to find the Pareto optimal scheduling of multiple DL applications in terms of the response time of each DL application and overall energy consumption under the given throughput constraints of DL applications. The proposed technique is verified with real DL networks running on two embedded devices, Galaxy S9 and HiKey970.INDEX TERMS Deep learning scheduling, genetic algorithm, heterogeneous processor, mobile device.

No abstract

ABSTRAcT. We examined sexual dimorphism in the craniodental traits of the raccoon dog Nyctereutes procyonoides from South Korea. Univariate comparisons of skull (cranium and mandible) and dental measurements revealed a small extent of sexual dimorphism in some measurements. The most indicative dimorphic measurements were the breadths of the upper and lower canines which were around 8% larger in male specimens on average. On the other hand, multivariate analyses using only skull traits showed slightly a clearer separation between sexes than those using only dental ones. This discrepancy may be derived from a higher variability in dental traits than in those of the skull. In conclusion, sexual dimorphism within N. procyonoides of South Korea is present, but was not so pronounced as for other local populations. However, measurements showing significant sexual dimorphism varied between different localities. This suggests that the selective forces acting upon craniodental morphology of each sex vary between populations of the species.

No abstract

The sequence of cranial suture closure among cervids is reported to be generally species-specific and highly conservative within species. On the other hand, it is known that intraspecific variation often exists to some extent in other mammalian taxa. Here we studied the cranial suture closures of Capreolus pygargus from Jeju Island and compared it with other cervid species. We found that the timing of the interparietal suture closure is highly variable within C. pygargus. Capreolus capreolus similarly shows intraspecific variation of the interparietal suture closure, whereas other cervid species studied to date do not show any intraspecific variation in the sequence of cranial suture closure. Such high intraspecific variation of the interparietal suture may be a derived character for Capreolus.