Shuffle federated learning is a promising federated learning paradigm that effectively balances the privacy and accuracy of the model. As the number of devices increases, the model accuracy produces a huge improvement, but the drastic increase of users can cause network problems, resulting in model sharing failure, longer global training time, and decreased model accuracy, so it is important to ensure that the local model completes the sharing. Thus, we propose a stochastically participating time-constrained shuffle federated learning model FedRtif. In FedRtif, the clients have the right to independently decide on random participation in training locally, and the shuffler also will perform the shuffling operation on the received local models within the constraint time to improve the model security, which alleviates the network congestion and ensures the safe sharing of local models. However, we find that the number of local models received at a fixed time is dynamically changing, which leads to reduced model security, and perturbation of all models also leads to reduced model accuracy. To address this challenge, we propose FedRtid, a dynamic timeconstrained adaptive shuffle federated learning model, which controls the amount of noise added through an adaptive K perturbation mechanism and employs adaptive constraints on the time to augment the number of models acquired by the shuffler to achieve an improvement in model accuracy and security. We conducted extensive experiments on three real datasets under independent and non-independent identically distributed data, and the results show that FedRtid has better model accuracy and privacy.
The arsenic (As) release from litter decomposition of As-hyperaccumulator (Pteris vittata L.) in mine areas poses an ecological risk for metal dispersion into the soil. However, the effect of atmospheric nitrogen (N) deposition on the litter decomposition of As-hyperaccumulator in the tailing mine area remains poorly understood. In this study, we conducted a microcosm experiment to investigate the As release during the decomposition of P. vittata litter under four gradients of N addition (0, 5, 10, and 20 mg N g−1). The N10 treatment (10 mg N g−1) enhanced As release from P. vittata litter by 1.2–2.6 folds compared to control. Furthermore, Streptomyces, Pantoea, and Curtobacterium were found to primarily affect the As release during the litter decomposition process. Additionally, N addition decreased the soil pH, subsequently increased the microbial biomass, as well as hydrolase activities (NAG) which regulated N release. Thereby, N addition increased the As release from P. vittata litter and then transferred to the soil. Moreover, this process caused a transformation of non-labile As fractions into labile forms, resulting in an increase of available As concentration by 13.02–20.16% within the soil after a 90-day incubation period. Our findings provide valuable insights into assessing the ecological risk associated with As release from the decomposition of P. vittata litter towards the soil, particularly under elevated atmospheric N deposition.
Abstract Aim We combined genetic sequence data and ecological niche modelling to resolve the impacts of past climatic fluctuations on the distribution, genetic diversification, and demographic dynamics of an East Asian montane bird, the green‐backed tit ( Parus monticolus ). Location East Asia. Methods Phylogenetic analyses were carried out using four mitochondrial fragments and seven nuclear loci from 161 birds sampled from 29 localities spanning the entire geographical range of the green‐backed tit. We used * beast to estimate the species tree and calculate divergence times. Extended Bayesian skyline plots were used to infer potential historical shifts in population size. We used MaxEnt to predict potential distributions during three periods: the present day, the Last Glacial Maximum and the Last Interglacial. Results The mitochondrial DNA (mt DNA ) gene tree showed strong support for three reciprocally monophyletic groups: a south‐western clade, a central clade and a Taiwanese clade. Taiwanese and Vietnamese samples had fixed differences at several nuclear loci, but the south‐western and central samples shared haplotypes at all nuclear loci. The mt DNA gene tree topology differed from the species tree topology. The species tree suggested sister relationships between Taiwanese and Vietnamese operational taxonomic units ( OTU s) and between south‐western and central OTU s. Diversification within the green‐backed tit was relatively recent, probably within the last 0.9 million years. Extended Bayesian skyline plots suggested rapid population expansion in the south‐western and central phylogroups after the Last Interglacial, and this result was consistent with ecological niche models. Main conclusions Our results suggest that genetic diversification within the green‐backed tit was affected by the later Pleistocene climate fluctuations. Ecological niche models indicated that the present‐day vegetation distribution was, in many ways, more similar to that of the Last Glacial Maximum than it was to that of the Last Interglacial. Continental populations of the green‐backed tit experienced unusual demographic and range expansion that is likely to have occurred during the cooling transition between the Last Interglacial and the Last Glacial Maximum. We found incongruence between the mt DNA gene tree and the species tree, which underscores the importance of using both mitochondrial and nuclear markers when estimating the evolutionary history of populations.
Most of the water-drive oil reservoirs in the western South China Sea had stepped in the middle and high water-cut stage. By the influence of reservoir heterogeneity, fault distribution, well pattern deployment and variation of reservoir flow parameters during long-term natural water drive and water flooding, the remaining oil distribution forecast is not accurate enough, increasing the difficulty of making effective adjustment and potential tapping measures. Through years of tackling key technical problems and field practice, the detailed characterization technique of the remaining oil in matured water drive reservoir was presented. Based on water displacement mechanism, variation of relative permeability curves are derived from Zhang’s water-drive characteristic curve during long-term water displacement. In addition, dynamic monitoring data matching was adopted to improve the forecast accuracy of the remaining oil distribution in water flooding oil reservoirs. By combination of flow field, remaining oil saturation field, and remaining oil reserves abundance, comprehensive characterization of water drive dynamic state was realized. The remaining oil enriched areas were quantitatively classified into four levels of potential regions, and corresponding adjustment and potential tapping measures were proposed. This technique had been successfully applied in the middle and high water-cut oilfields in the western South China Sea, with remarkable estimated incremental oil production of approximately 204,000 m3.
According to the nonlinear characteristic of solar cell output, the TMPPT of solar cell is studied and the tracking control strategy applying varying voltage step disturbing is presented. Combining with the charging and discharging curve of battery, the favorable management study on it is done. The software and hardware of photovoltaic charging system is designed based on the microprocessor JK3, and the some experiments on the TMPPT of solar cell are done with solar cell array simulator. The above control strategy is proved to be effective and better by experiment results.
The biogeochemical effect of birds can influence the paedogenic processes in the Antarctica soils. The content of nutrients (phosphorus, potassium, nitrate and ammonium) and the total carbon and nitrogen contents in soils were determined at the nesting and feeding sites of flying and non-flying birds in the Fildes Peninsula (King George Island, West Antarctica). The results showed that the difference between soils affected by birds (both flying and non-flying) and the background soils was observed with the higher content of biogenic elements (e.g. C, N, P, K) and fine-earth content, variable soil pH values, more expressed soil structure and specific morphology of the uppermost organogenic and organo-mineral horizons. Additional chemical indices are suggested for diagnostics of current and recent effect of birds on paedoenvironments.
For the data streams retrieved from the various sensors in energy cyber physical system, an anomaly detection algorithm based on time series extremum was proposed. Firstly, for a single data stream, the algorithm identifies the extreme points of the data stream in the sliding window, and marks off the abnormal sub-sequences according to these points. Then the eigenvalues of the sub-sequences are calculated to establish the feature space. Several anomaly measures are defined, so the sub-sequences can be sorted by the measures in the feature space. Finally, based on the correlation between the sensors points, the anomaly points on each work section are detected. The algorithm can make up for the limitation of other point anomaly detection algorithms by detecting the abnormal sub-sequences, and reduces the probability of false alarm and missing alarm in data stream analysis. Experiments show that the algorithm is effective and feasible.
'/%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)