Search

Self-supervised learning (SSL) has emerged as a promising approach for learning representations from unlabeled data. Momentum-based contrastive frameworks such as MoCo-v3 have shown remarkable success among the many SSL methods proposed in recent years. However, a significant gap in encoder representation exists between the online encoder (student) and the momentum encoder (teacher) in these frameworks, limiting the performance on downstream tasks. We identify this gap as a bottleneck often overlooked in existing frameworks and propose “residual momentum” that explicitly reduces the gap during training to encourage the student to learn representations closer to the teacher’s. We also reveal that a similar technique, knowledge distillation (KD), to reduce the distribution gap with cross-entropy-based loss in supervised learning is useless in the SSL context and demonstrate that the intra-representation gap measured by cosine similarity is crucial for EMA-based SSLs. Extensive experiments on different benchmark datasets and architectures demonstrate the superiority of our method compared to state-of-the-art contrastive learning baselines. Specifically, our method outperforms MoCo-v3 0.7% top-1 in ImageNet, 2.82% on CIFAR-100, 1.8% AP, and 3.0% AP75 on VOC detection pre-trained on the COCO dataset; it also improves DenseCL with 0.5% AP (800ep) and 0.6% AP75 (1600ep). Our work highlights the importance of reducing the teacher-student intra-gap in momentum-based contrastive learning frameworks and provides a practical solution for improving the quality of learned representations.

For over 30 years, extensive research has found corroborating evidence that past winners continue to yield higher returns than past losers. This momentum effect is robust across various asset classes and across the globe and presents perhaps the most pervasive contradiction of the efficient market hypothesis. This article reviews three strands of literature on momentum. First, I outline the construction of momentum strategies, emphasizing improvements and alternatives such as time-series momentum, residual momentum, and risk-managed momentum. Second, I summarize the most prominent behavioral-based and risk-based explanations for the origin of momentum. Finally, I present in detail the findings on commonality in stock momentum, namely on industry and factor momentum. [ABSTRACT FROM AUTHOR]

We find that the conventional time-series momentum strategy can be revised in two ways to deliver enhanced excess return performance that are reminiscent of the advancements for the cross-sectional momentum strategy. First, we explore the applicability of the residual return for time-series momentum and find that it opens a door for constructing new improved time-series momentum strategies. Second, we examine the pertinence of the auto-covariance property of returns, which is the defining characteristic of time-series momentum, for asset allocation which has almost universally been identified with the equal-weighted one in the literature. We discover that consistent and persistent performance enhancements are earned by both approaches for the industry portfolios. [ABSTRACT FROM AUTHOR]

In the tangential combustion of a pulverized coal boiler, the swirl number is the key design index for evaluating the flue-gas temperature deviation at both sides of the furnace outlet. However, the current calculation method ignores the effect of the distances of the burner nozzles to the furnace outlet, and thus for large-scale utility boilers, the swirl number is significantly large. As such, the calculated result does not coincide with the actual situation. In this study, the calculation method of the swirl number was corrected by considering the effect of the distances of the burner nozzles to the furnace outlet. In addition, the corrected calculation method was verified through actual operation. The obtained results prove that the corrected calculation method is more accurate and reliable. In addition, the study indicates that for the special four-wall tangential combustion with a large tangential circle, the separated over fire (SOFA) air must have a four-corner tangential arrangement for the rotating airflow is still strong at the bottom of the platen and without other ways to reduce it. [ABSTRACT FROM AUTHOR]

• Residual momentum strategy yields significant profits in the Chinese equity market. • Residual momentum returns cannot be explained by four popular asset pricing models. • Residual momentum profits do not reverse in the long run. • The residual momentum factor is priced in the cross-section of stock returns. In this paper, I find that sorting stocks into portfolios based on their residual, as opposed to raw past returns, generates significant profits in the Chinese equity market and cannot be subsumed by the well-established factor models. Moreover, the residual momentum profits do not reverse in the long run (up to three years), supporting the investor underreaction hypothesis. Further analysis reveals that residual momentum is priced in the cross-section of stock returns whereas the Carhart (1997) momentum factor is found to be redundant for describing average stock returns. [ABSTRACT FROM AUTHOR]

Highlights • A robust landmark-based algorithm for finding the geometric median (group average) of a group of shapes with heavy outliers is proposed. • The underlying Log and Exp maps of the proposed algorithm belong to a class of efficient geodesic shooting algorithms for template matching. • Once the group average is found, the Hamiltonian metric computed between the group average and each member of the group can be used for clustering analysis. • The proposed algorithms, as a tool of feature extraction, extract information about momentum at each landmark during the process of finding the geometric median. • The local momenta found by the proposed algorithm are useful for some classification purposes that cannot be achieved by using the landmark locations. Abstract We introduce a class of mathematical algorithms with the aim of establishing a framework of finding a group average and extracting prominent features in a group of landmark represented shapes or image templates. A group average is an estimator that is said to best represent the common features of the group being studied. The proposed algorithms, as a tool of feature extraction, extract information about momentum at each landmark through the process of template matching. Once the convergence criterion is satisfied numerically, the algorithms produce a group average and a local coordinate system for each member of the observing group, in terms of the residual momentum. We present several examples to illustrate the use of the proposed algorithms for finding a group average. Using the metrics computed between the group average and each member of the group, we successfully run a cluster analysis for datasets that contain a heavy percentage of outliers. Finally, we apply the collected residual momenta computed in the proposed algorithms in some statistical methods to demonstrate a potential application of the algorithms for detecting structure abnormality. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Management & Business Research (2521-4306) is the property of Chinese Management Association and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

We offer a new type of momentum strategy -- the volatility-adjusted residual momentum (VARMOM) -- which is based on average past residuals scaled with their volatility. We demonstrate its application for international asset allocation within 51 country indexes and 888 industry portfolios from developed and emerging markets. The VARMOM trading strategy notably outperforms and subsumes a standard momentum strategy, delivering Sharpe ratios that are two to three times higher. The VARMOM is particularly strong across portfolios characterised by high limits to arbitrage and following bull markets, supporting the behavioural explanation of momentum. The results are robust to alternative portfolio construction methods as well as the inclusion of trading costs and control variables. They are also valid for several subperiods and subsamples. [ABSTRACT FROM AUTHOR]