Search

Although agenda-setting dynamics in the European Union are a well-studied phenomenon, there is a gap in the literature between the current focus on issue attention dynamics (a policy input) and the impact attention dynamics have on policy outputs. This study examines how the rotating Council Presidency’s stated policy goals for their term in office affect EU decision-making efficiency. We show that the salience the rotating chair of the Council attaches to a given policy area affects legislative efficiency in that policy area. We also demonstrate how this effect is conditioned on government effectiveness in the state of the Presidency in question, and is independent of Commission policy priorities. This suggests that the Presidency can drive EU policy outputs and push for its priorities when it holds the chair, but that its ability to do so is contingent on the domestic organisation and effectiveness of the state.

This article unveils the policy agenda of the European Central Bank (ECB) Governing Council as found in the speeches that Governing Council Members gave between 1999 and 2018. Using a dynamic topic‐modeling approached based on non‐negative matrix factorization, we demonstrate how the issues discussed by ECB Governing Council members have evolved over time, and how the general punctuation hypothesis (Jones, B. D. & Baumgartner, F. R. (2005). The politics of attention: How government prioritizes problems. University of Chicago Press) sheds light on what drives this process. We find that unlike policy outputs from many other policymaking systems, ECB communications evolve in a proportional manner. We attribute this finding to the information‐processing capacities of the bank. Our findings speak to the literatures on central bank communications, the evolution of policy agendas, and the application of topic models to speech texts. Science Foundation Ireland Insight Research Centre Update issue date during checkdate report - AC

Multi-task learning with an unbalanced data distribution skews model learning towards high resource tasks, especially when model capacity is fixed and fully shared across all tasks. Sparse scaling architectures, such as BASELayers, provide flexible mechanisms for different tasks to have a variable number of parameters, which can be useful to counterbalance skewed data distributions. We find that that sparse architectures for multilingual machine translation can perform poorly out of the box, and propose two straightforward techniques to mitigate this - a temperature heating mechanism and dense pre-training. Overall, these methods improve performance on two multilingual translation benchmarks compared to standard BASELayers and Dense scaling baselines, and in combination, more than 2x model convergence speed.

Multilingual neural machine translation has shown the capability of directly translating between language pairs unseen in training, i.e. zero-shot translation. Despite being conceptually attractive, it often suffers from low output quality. The difficulty of generalizing to new translation directions suggests the model representations are highly specific to those language pairs seen in training. We demonstrate that a main factor causing the language-specific representations is the positional correspondence to input tokens. We show that this can be easily alleviated by removing residual connections in an encoder layer. With this modification, we gain up to 18.5 BLEU points on zero-shot translation while retaining quality on supervised directions. The improvements are particularly prominent between related languages, where our proposed model outperforms pivot-based translation. Moreover, our approach allows easy integration of new languages, which substantially expands translation coverage. By thorough inspections of the hidden layer outputs, we show that our approach indeed leads to more language-independent representations., ACL 2021

Recent work in multilingual translation advances translation quality surpassing bilingual baselines using deep transformer models with increased capacity. However, the extra latency and memory costs introduced by this approach may make it unacceptable for efficiency-constrained applications. It has recently been shown for bilingual translation that using a deep encoder and shallow decoder (DESD) can reduce inference latency while maintaining translation quality, so we study similar speed-accuracy trade-offs for multilingual translation. We find that for many-to-one translation we can indeed increase decoder speed without sacrificing quality using this approach, but for one-to-many translation, shallow decoders cause a clear quality drop. To ameliorate this drop, we propose a deep encoder with multiple shallow decoders (DEMSD) where each shallow decoder is responsible for a disjoint subset of target languages. Specifically, the DEMSD model with 2-layer decoders is able to obtain a 1.8x speedup on average compared to a standard transformer model with no drop in translation quality., Comment: EACL 2021

In this paper, the nature of thermally upgraded Kraft electrical insulating papers and their characterization by measurement of nitrogen content is discussed. A historical context and explanation of the basics of the various chemical methods used to measure nitrogen content in electrical papers is presented. The Kjeldahl and Dumas methods for quantifying nitrogen content were compared using six insulating papers in different laboratories in an effort to correlate the results obtained with both techniques. The results provide an insight into the viability of the Dumas combustion method as a proposed industry standard test for measuring the nitrogen content of paper. Moreover, data on the accuracy, reproducibility, and repeatability of the Dumas method is provided for the analysis of thermally upgraded electrical papers.

The ability to amend legislative proposals introduced by the Commission is central to legislative process in the European Union. Despite this, very few attempts have been made to capture and explain such amendments. This study addresses this gap in the literature by considering the changes between the Commission’s proposals and the final legislative outcome passed by the European Union. It does so by implementing minimum edit distance algorithms to measure changes between legislative proposals and outcomes. The findings suggest that legislative amendments are determined by the formal and informal institutional structures in which negotiations take place and characteristics of the proposal itself. Our conclusions contribute to the ongoing debate on the nature and distribution of legislative powers in the European Union.