The UK Research Assessment Exercise and the narrowing of UK economics

Abstract

Federated learning (FL) has enabled distributed learning of a model across multiple clients in a privacy-preserving manner. One of the main challenges of FL is to accommodate clients with varying hardware capacities; clients have differing compute and memory requirements. To tackle this challenge, recent state-of-the-art approaches leverage the use of early exits. Nonetheless, these approaches fall short of mitigating the challenges of joint learning multiple exit classifiers, often relying on hand-picked heuristic solutions for knowledge distillation among classifiers and/or utilizing additional layers for weaker classifiers. In this work, instead of utilizing multiple classifiers, we propose a recurrent early exit approach named ReeFL that fuses features from different sub-models into a single shared classifier. Specifically, we use a transformer-based earlyexit module shared among sub-models to i) better exploit multi-layer feature representations for task-specific prediction and ii) modulate the feature representation of the backbone model for subsequent predictions. We additionally present a per-client self-distillation approach where the best sub-model is automatically selected as the teacher of the other sub-models at each client. Our experiments on standard image and speech classification benchmarks across various emerging federated fine-tuning baselines demonstrate ReeFL's effectiveness over previous works.

An investigation into the role played by research assessment in the socio-geographic fragmentation of undergraduate economics education in the UK

“…Research groups that perform below their universities' expectations may be candidates to be pulled out of the next E&E UOA (Johnston and Reeves 2017). One of the consequences of performing less well than expected may be a cut in funding or, very likely, no funding at all (Lee et al 2013;Sayer 2015). Unfortunately, it is very difficult for an outside observer to know the expected rating for any given university.…”

“…In E&E, publications in elite journals (e.g. the so-called Diamond list) are essential for a high rating to guarantee funding for future research (Lee et al 2013). It is believed by some that the Diamond list journals are open only to those with specialist training in econometrics and work against researchers working in areas that do not excel in econometric techniques (Lodewijks and Stokes 2014).…”

Gekko and black swans: Finance theory in UK undergraduate curricula

“…It would be wrong, however, to state that finance/economics theory has been an inert and concretised subject. Although mainstream economics has been in existence for a number of years, heterodox economics has gained some support in recent times (Lee et al, 2013;Lee, 2012;Maio, 2013). Lee (2012, p. 339) refers to it as a "particular group of contemporary theories aimed at explaining the social provisioning process, to economic policies recommendations predicated on the theories, and to a community of economists engaged in this theoretical and applied scientific activity.…”

“…1) and as key proponents of heterodox economics, Lee et al (2013) discuss how UK neoclassical economists are intolerant of any other paradigm.…”

“…Although, imperfect markets and information exist in the curriculum, there appear to be optimal ways around them (Giacalone & Thompson, 2006). In short, the focus is on use of neoclassical theory (Findlay, 1987;Merton, 2002);Miller, 2000) and on excluding other pluralistic/sociological theories (Lee, 2012;Lee, Pham, & Gu, 2013;Gendron & Smith-Lacroix, 2013). Practice is based on "techniques and tools" (see for e.g.…”