Minimal Gated Unit for Recurrent Neural Networks Zhou, Guo-Bing; Wu, Jianxin; Zhang, Chen-Lin ...
International journal of automation and computing,
06/2016, Letnik:
13, Številka:
3
Journal Article
Recenzirano
Odprti dostop
Recurrent neural networks （RNN） have been very successful in handling sequence data. However, understanding RNN and finding the best practices for RNN learning is a difficult task, partly because ...there are many competing and complex hidden units, such as the long short-term memory （LSTM） and the gated recurrent unit （GRU）. We propose a gated unit for RNN, named as minimal gated unit （MCU）, since it only contains one gate, which is a minimal design among all gated hidden units. The design of MCU benefits from evaluation results on LSTM and GRU in the literature. Experiments on various sequence data show that MCU has comparable accuracy with GRU, but has a simpler structure, fewer parameters, and faster training. Hence, MGU is suitable in RNN＇s applications. Its simple architecture also means that it is easier to evaluate and tune, and in principle it is easier to study MGU＇s properties theoretically and empirically.
Context
Existing landscape sustainability studies mostly follow design-based, practical approaches, emphasizing “linking knowledge to action;” while the problem-driven, diagnostic approach that ...emphasizes “understanding human–environment interactions” remains underused.
Objectives
To apply the problem-driven, diagnostic approach to a widespread and controversial landscape-relevant problem: farmland loss associated with the long-term, global trend of urbanization.
Methods
We comparatively analyzed the farmland loss and landscape dynamics in the core and peripheral areas of rapidly urbanizing Tongling City in China, by using 30 m-grid land use/cover data in 2000, 2005, 2010, and 2015. To understand the temporal trends and abnormalities of farmland loss, we identified probable regime shifts of socioeconomic and agrifood dynamics by changepoint detection using relevant data from 2000 to 2015.
Results
Changes of the whole landscape were dominated by persistent aggressive land gains of developed lands and increasing land losses of farmland and densely/moderately vegetated land. The ratio of farmland loss to total land conversion (8–77%) decreased in general during 2000–2015, with urban encroachment of farmland accounting for 14.72–74.36% of total farmland loss. Farmland had become smaller in patch size, less regular in patch shape, and more isolated between patches since 2005, indicating farmland deintensification. Three temporal abnormalities of farmland change were identified, which were associated with abrupt socioeconomic and agrifood dynamics affecting farmland demand–supply.
Conclusions
Farmland loss is more of an economic issue than a resource issue, and focusing on urban expansion is often inadequate. Policy efforts are needed to address the real problem of social injustice in mandatory farmland preservation practices.
2D amorphous transition metal oxides (a‐TMOs) heterojunctions that have the synergistic effects of interface (efficiently promoting the separation of electron−hole pairs) and amorphous nature ...(abundant defects and dangling bonds) have attracted substantial interest as compelling photocatalysts for solar energy conversion. Strategies to facilely construct a‐TMOs‐based 2D/2D heterojunctions is still a big challenge due to the difficulty of preparing individual amorphous counterparts. A generalized synthesis strategy based on supramolecular self‐assembly for bottom–up growth of a‐TMOs‐based 2D heterojunctions is reported, by taking 2D/2D g‐C3N4 (CN)/a‐TMOs heterojunction as a proof‐of‐concept. This strategy primarily depends on controlling the cooperation of the growth of supramolecular precursor and the coordinated covalent bonds arising from the tendency of metal ions to attain the stable configuration of electrons, which is independent on the intrinsic character of individual metal ion, indicating it is universally applicable. As a demonstration, the structure, physical properties, and photocatalytic water‐splitting performance of CN/a‐ZnO heterojunction are systematically studied. The optimized 2D/2D CN/a‐ZnO exhibits enhanced photocatalytic performance, the hydrogen (432.6 µmol h−1 g−1) and oxygen (532.4 µmol h−1 g−1) evolution rate are 15.5 and 12.2 times than bulk CN, respectively. This synthetic strategy is useful to construct 2D a‐TMOs nanomaterials for applications in energy‐related areas and beyond.
A generalized synthesis strategy for the bottom–up growth of amorphous transition metal oxides (a‐TMOs) 2D/2D heterojunctions with large contact area via covalent interfacial interaction is provided. A number of 2D/2D CN/a‐TMOs heterojunctions, such as CN/a‐FeOx, CN/a‐CuOx, CN/a‐MnOx, CN/a‐CoOx, and CN/a‐ZnO are successfully synthesized, which has a large and perfect order−disorder interface. Particularly, 2D/2D CN/a‐ZnO heterojunction exhibit boosted photocatalytic hydrogen and oxygen evolution.
Urbanization is a long-term global trend critical for shaping human-Earth sustainability in the Anthropocene. In past decades, much progress has been made in researching urban sustainability, ...particularly global assessments of the big picture and case studies of individual cities. Here we examine the world's top 100 urban agglomerations (UAs) in terms of size—that rank high on sustainability agendas and cover 28% of the global UA area—regarding four broadly concerned challenges: population shrinkage, slum development, greenness loss, and heat exposure. Instead of merely focusing on global/regional “averages” or individual cases, we take one step further to identify the “anomalies” of urban sustainability among the 100 UAs for each dimension and on the whole as multi-dimensional coupled infrastructure systems. Results show: (1) urban population of the 100 UAs increased by 36% during 2000–2020; (2) urban slums occurred in 85% of 34 examined UAs in the Global South; (3) urban greenness declined in the 100 UAs by 8% during 2000–2019; and (4) 79% of the 100 UAs were projected to have less than 30 EHDs per year during 2021–2030. Our findings provide global baselines for place-based problem-driven policymaking for the examined UAs and suggest improving urban green infrastructure as their top policy imperative. Our findings point to a critical research gap in the urban sustainability literature: Studying sustainability transitions of the “abnormally” sustainable UAs identified in this study that had exceptional performances on the four examined sustainability dimensions, e.g., Beijing of China and Milan of Italy.
•The world's top 100 UAs were identified, accounting for 28% of the global UA area.•The general trends/patterns of four population-environment dynamics were quantified.•A policy imperative is to meet the rapidly growing population's infrastructure needs.•Abnormally sustainable UAs were highlighted which point to a new research avenue.
The resistance of breast cancer cells to drugs is a major obstacle to effective cancer chemotherapy. Here, we study the function mechanisms of long non‐coding RNA XIST in chemoresistance of breast ...cancer to doxorubicin. We examined the 50% inhibitive concentration of doxorubicin to MDA‐MB‐231 and MDA‐MB‐231/ADM cells, showing that the doxorubicin resistance of MDA‐MB‐231/ADM cells was much higher than MDA‐MB‐231 cells. The gene or protein expression of XIST and ANLN were also higher in MDA‐MB‐231/ADM cells than that in MDA‐MB‐231 cells. Moreover, XIST overexpression promoted cell proliferation and inhibited apoptosis of doxorubicin‐treated MDA‐MB‐231 cells by promoting ANLN expression. XIST silencing inhibited cell proliferation and promoted apoptosis of doxorubicin‐treated MDA‐MB‐231/ADM cells by inhibiting ANLN expression. Luciferase reporter assay showed that XIST functioned as a competing endogenous RNA to repress miR‐200c‐3p, which controlled its downstream target ANLN. In conclusion, these data reveal that XIST promotes chemoresistance of breast cancer cells to doxorubicin by sponging miR‐200c‐3p to upregulate ANLN. This work explores the relationship between lncRNA XIST and doxorubicin resistance in breast cancer cells and highlights a novel therapeutic target for the treatment of breast cancer.
•We conducted a systematic review of 333 papers by focusing on two core concepts.•Sustainability studies of landscapes have accelerated in number since 2004–2006.•The ecological and ...management/planning dimensions have been more studied.•The sociocultural and theoretical dimensions have gained growing interest recently.•Two challenges to and new directions for future studies were discussed.
Landscape scientists have increasingly studied sustainability during the past three decades, with a plurality of perspectives and methods. However, a comprehensive review of the relevant literature is still lacking. Two concepts capture the core of these studies: sustainable landscapes (SL) and landscape sustainability (LS). Here we present a bibliometric analysis of 333 English papers published in SCI journals (i.e., indexed by Web of Science) during 1990–2017, whose titles, abstracts or keywords contain SL, LS, or both. Using multiple methods, including change-point detection, text and topic mining, and citation analysis, we found: (1) Sustainability studies of landscapes have entered a rapid growth phase since 2004–2006, as determined statistically by the annual publications and citations; (2) There have been many more studies focusing on the ecological and practical dimensions than sociocultural and theoretical dimensions; and (3) Influenced by advances in sustainability research in a broader science context, studies of SL and LS have become increasingly holistic, transdisciplinary, and normative. Our findings suggest that, to further advance landscape sustainability science, scholars need to be more explicit about the underlying sustainability perspective and associated scales of key terms (e.g., SL, LS, and ecosystem services) in specific studies, and need to put more emphasis on place-based case studies of various landscape unsustainability syndromes by integrating the sustainability science-based social-ecological systems and transdisciplinary perspectives.
Context
In China, large cities are often held to blame for the drastic loss of high-quality farmland. However, others propose that the development of large cities should be prioritized for their ...scale economy in saving farmland. A policy debate remains regarding which urbanization mode—developing a few large or many small cities—can minimize the impacts of urban expansion on farmland loss.
Objective
The present study aims to clarify the controversy by quantifying the quantity and quality of urbanization-encroached farmland (UEF) in different sizes of cities.
Methods
We proposed two indexes to quantify the UEF among different sizes of cities during 2000–2010: “the UEF area per urban population growth” and “the average potential yield of UEF”. Furthermore, we proposed a spatial prioritization matrix to facilitate place-based policy design for UEF management.
Results
The super large cities (SLC) consumed the least UEF area per urban population growth and also the lowest average potential yield of UEF. However, there were large variations within city categories in both the quality and quantity dimensions. Farmland loss to urban expansion is determined by a city’s biophysical and socio-political conditions.
Conclusions
Our analyses show that urban growth of SLC consumed smaller quantity of farmland when accommodating a certain amount of urban population. Containing the development of large cities in the name of food security is ineffective, which will cause double failures in farmland preservation and urban development. As the impacts of urban expansion on farmland loss are highly place-variant and urban sustainability is more than just food security, place-based policies following the landscape sustainability perspective are needed to address urban encroachment of farmland.
Abstract
To address future environmental change and consequent social vulnerability, a better understanding of future population (FPOP) dynamics is critical. In this regard, notable progress has been ...made in producing FPOP projections that are consistent with the Shared Socioeconomic Pathways (SSPs) at low resolutions for the globe and high resolutions for specific regions. Building on existing endeavors, here we contribute a new set of 1 km SSP-consistent global population projections (FPOP in short for the dataset) under a machine learning framework. Our approach incorporates a recently available SSP-consistent global built-up land dataset under the Coupled Model Intercomparison Project 6, with the aim to address the misestimation of future built-up land dynamics underlying existing datasets of future global population projections. We show that the overall accuracy of our FPOP outperforms five existing datasets at multiple scales and especially in densely-populated areas (e.g. cities and towns). Followingly, FPOP-based assessments of future global population dynamics suggest a similar trend by population density and a spatial Matthew effect of regional population centralization. Furthermore, FPOP-based estimates of global heat exposure are around 300 billion person-days in 2020 under four SSP-Representative Concentration Pathway (RCPs), which by 2100 could increase to as low as 516 billion person-days under SSP5-RCP4.5 and as high as 1626 billion person-days under SSP3-RCP8.5—with Asia and Africa contributing 64%–68% and 21%–25%, respectively. While our results shed lights on proactive policy interventions for addressing future global heat hazard, FPOP will enable future-oriented assessments of a wide range of environmental hazards, e.g. hurricanes, droughts, and flooding.
PDF
