No abstract

No abstract

Although humans can easily identify the object of interest from groups of examples using group-level labels, most of the existing machine learning algorithms can only learn from individually labeled examples. Multi-instance learning (MIL) is a type of weakly supervised learning that deals with objects represented as groups of instances, and is theoretically capable of predicting instance labels from group-level supervision. Unfortunately, most existing MIL algorithms focus on improving the performances of group label predictions and cannot be used to accurately predict instance labels. In this work, we propose the TargetedMIL algorithm, which learns semantically meaningful representations that can be interpreted as causal to the object of interest. Utilizing the inferred representations, TargetedMIL excels at instance label predictions from group-level labels. Qualitative and quantitative evaluations on various datasets demonstrate the effectiveness of TargetedMIL.

Background Grazing exclusion is a common grassland management strategy for restoring degraded grasslands. Its effectiveness on optimizing plant species community, increasing vegetation diversity and biomass, improving soil fertility, has been widely documented in literatures. However, little is known on the responses of the absolute abundance and the ecological functions of soil bacterial community to long-term grazing exclusion. Result In this study, the absolute abundance, diversity, and ecological functions of soil bacterial community were determined by the high-throughput absolute quantitative sequencing technology on a long-term grazing exclusion (40 years, GE) area and three free grazing areas (FGs) within a Leymus chinensis steppe of Inner Mongolia, China, and analyzed the driving forces leading to the variations in soil bacterial community and functions. Our results showed that there was significantly higher soil bacterial abundance in the GE than the FGs along with corresponding variations in vegetation and soil properties. With the decrease of vegetation aboveground biomass, the absolute abundance of soil bacterial community also decreased. Among the phyla of the soil bacterial communities, the relative abundances of Chloroflexi and Firmicutes phyla were especially lower, and that of Verrucomicrobia phylum was higher in the GE than the FGs; the absolute abundances of Euryarchaeota and Microgenomates phyla were especially higher in the GE than the FGs. Conclusions This study suggested that long-term grazing exclusion significantly increased the absolute abundance, changed soil bacterial composition, and especially enhanced bacterial motility and chemotaxis. In particular, soil organic matter was the important agent to influence and connect vegetation and soil. This work will enrich our understanding of the responses of absolute abundance, diversity, and function of the soil bacterial community to long-term grazing exclusion, and help the evaluation of grassland degradation degree and restoration strategy effectiveness.