Recent LSS-based multi-view 3D object detection has made tremendous progress, by processing the features in Brid-Eye-View (BEV) via the convolutional detector. However, the typical convolution ignores the radial symmetry of the BEV features and increases the difficulty of the detector optimization. To preserve the inherent property of the BEV features and ease the optimization, we propose an azimuth-equivariant convolution (AeConv) and an azimuth-equivariant anchor. The sampling grid of AeConv is always in the radial direction, thus it can learn azimuth-invariant BEV features. The proposed anchor enables the detection head to learn predicting azimuth-irrelevant targets. In addition, we introduce a camera-decoupled virtual depth to unify the depth prediction for the images with different camera intrinsic parameters. The resultant detector is dubbed Azimith-equivariant Detector (AeDet). Extensive experiments are conducted on nuScenes, and AeDet achieves a 62.0% NDS, surpassing the recent multi-view 3D object detectors such as PETRv2 and BEVDepth by a large margin. Project page: https://fcjian.github.io/aedet.
Behavioral experiments on the Ultimatum game have shown that we human beings have a remarkable preference in fair play, contradicting the predictions by the game theory. Most of the existing models seeking for explanations, however, strictly follow the assumption of Homo economicus in orthodox economics that people are self-interested and fully rational to maximize their earnings. Here we relax this assumption by allowing that people probabilistically choose to be ‘good Samaritans’, acting as fair players from time to time. For well-mixed and homogeneously structured populations, we numerically show that as this probability increases the level of fairness undergoes from the low scenario abruptly to the full fairness state, where occasional fair behaviors are sufficient to drive the whole population to behave in the half–half split manner. We also develop a mean-field theory, which correctly reproduces the first-order phase transition and points out the reason. Heterogeneously structured populations, however, display continuous fairness transition; surprisingly, very few hub nodes acting as fair players are able to entrain the whole population to the full fairness state. Our results thus reveal the unexpected strength of ‘good Samaritans’, which may constitute a new explanation for the emergence of fairness in our society.
This paper constructs a new firm-level measure of ESG disclosure quality(ESGDQ) defined as the proportion of non-missing items in the LSEG (Refinitiv)ESG scoring template. ESGDQ captures the completeness of a firm’s publiclydisclosed ESG information. Using a panel of U.S. listed firms from 2009 to 2022,we validate ESGDQ against established ESG proxies and find it negatively asso-ciated with information asymmetry. Higher ESGDQ is associated with improvedstock liquidity after controlling for firm and industry-by-year fixed effects andthis holds across ESG pillars and categories. Heterogeneity tests show that thisassociation is stronger when retail investor attention and analyst coverage arelower. Finally, an event study further finds that U.S. firms with greater EUexposure exhibit higher environmental disclosure quality prior to 2020 while less-exposed firms converge toward their level afterward, a trend accompanied bycorresponding improvements in stock liquidity.
Cette thèse s’interroge sur les notions d’aléa et de régularité des variations boursières. Nous démontrons sur le plan théorique, la compatibilité des principales théories financières (cf. efficience informationnelle, finance comportementale et approche conventionnaliste) avec l'impossibilité de battre la stratégie «buy and hold». Cette impossibilité est confirmée par les études statistiques dans la mesure où les régularités identifiées dans les séries financières ne permettent pas de prédire le sens des variations futures. Les modèles économétriques disponibles à présent offrent souvent un «hit score» insuffisant (<60%) pour réussir des tentatives fructueuses de «market timing». Une contribution de ce travail se trouve dans l'introduction du concept de complexité algorithmique en finance. Une approche générale est proposée pour estimer la «complexité de Kolmogorov» des séries de rentabilités: après un processus «discrétisation-effacement», des algorithmes de compression sans perte sont utilisés pour détecter des structures régulières qui ne sont pas toujours visibles aux yeux des tests statistiques. En étudiant le degré d'aléa des principaux marchés internationaux à une fréquence «tick-by-tick», on constate une complexité plus élevée pour Euronext-Paris que pour le NYSE et le NASDAQ. Nous expliquons ce résultat par une auto-corrélation plus élevée des volatilités inter-journalières aux Etats-Unis qu'en France. L'inefficacité de «market timing» étant soutenue aussi bien par les théories financières que par les observations empiriques, nous définissons la notion de «battre le marché» dans ce sens spécifique avec un modèle mathématique qui s'inscrit dans le cadre de la calculabilité.