Search

Canadian environmental governance formally began in the 1970s with an ambitious vision critical of the status quo. Until the Justin Trudeau government from 2015, this vision had been largely eroded when compared to the transformative environmental ideas, ambition, and efforts initially put forth. Drawing on elite interviews in Canadian environmental politics and an articulation of the dominant system, we develop and demonstrate a novel explanation for Canada's systematic failure to act more ambitiously on the environment. We argue that this failure is the result of progressive selection of Canadian environmental governance in relation to the dominant system at key selection moments. It is at those moments when the environment is least prioritized and the economy most urgently needs attention that environmental governance suffers the most. The paper concludes by emphasizing how features of the dominant political-economic system not typically thought of as environmental can have systematic environmental impacts. [ABSTRACT FROM AUTHOR]

• This paper proposes a sampling framework for clustering algorithm. • It inherits from both progressive sampling and stratification concepts. • New representatives are selected if they improve the already selected set of samples. • The framework is independent on the sampling f as well as on the clustering one. • Performance is assessed using synthetic and real data. Clustering algorithms become more and more sophisticated to cope with large data sets of increasing complexity. Sampling selection methods are likely to provide an interesting alternative as they can reduce memory requirements, and reduce execution time. Many sampling algorithms for clustering are efficient but they each have their own limitations with large data sets. In this paper, we introduce a sampling framework for clustering algorithms that inherits from both progressive sampling and stratification concepts. Driven by two parameters, the iterative process consists in managing representatives of independent strata that carry similar statistical information regarding the clustering objective. At each iteration, the candidate representatives of the incoming stratum are examined. The interesting feature of the framework stems from the idea of selecting new representatives of the incoming stratum only if they improve the representation quality of the already selected set of samples. The algorithm stops when new representatives are no longer needed, which is likely to happen without examining the whole data set. The tests conducted on synthetic and real world datasets proved that the progressive sampling framework yielded similar results to the sampling algorithm applied to the whole set in a low computational time. In comparison with progressive sampling techniques, using the proposed framework enables smaller sampling sets to be used without loss of accuracy. [ABSTRACT FROM AUTHOR]

In recent years, many Unsupervised Domain Adaptation (UDA) methods have been proposed to tackle the domain shift problem. Most existing UDA methods are derived for Close Set Domain Adaptation (CSDA) in which source and target domains are assumed to share the same label space. However, target domain may contain unknown class different from the known ones in the source domain in practice, i.e., Open Set Domain Adaptation (OSDA). Due to the presence of unknown class, aligning the whole distribution of the source and target domain for OSDA as in the previous methods will lead to negative transfer. Existing methods developed for OSDA attempt to assign smaller weights to target samples of unknown class. Despite promising performance achieved by existing methods, the samples of the unknown class are still used for distribution alignment, which makes the model suffer from the risk of negative transfer. Instead of reweighting, this paper presents a novel method namely Thresholded Domain Adversarial Network (ThDAN), which progressively selects transferable target samples for distribution alignment. Based on the fact that samples from the known classes must be more transferable than target samples of the unknown one, we derive a criterion to quantify the transferability by constructing classifiers to categorize known classes and to discriminate unknown class. In ThDAN, an adaptive threshold is calculated by averaging transferability scores of source domain samples to select target samples for training. The threshold is tweaked progressively during the training process so that more and more target samples from the known classes can be correctly selected for adversarial training. Extensive experiments show that the proposed method outperforms state-of-the-art domain adaptation and open set recognition approaches on benchmarks. [ABSTRACT FROM AUTHOR]