PurposeThis study aimed to investigate the construct of external visual imagery (EVI) vs. internal visual imagery (IVI) by comparing the athletes' imagery ability with their levels of skill and types of sports.MethodsSeventy-two young athletes in open (n = 45) or closed (n = 27) sports and with different skill levels completed 2 custom-designed tasks. The EVI task involved the subject generating and visualizing the rotated images of different body parts, whereas the IVI task involved the subject visualizing himself or herself performing specific movements.ResultsThe significant Skill-Level × Sport Type interactions for the EVI task revealed that participants who specialized in open sports and had higher skill-levels had a higher accuracy rate as compared to the other subgroups. For the IVI task, the differences between the groups were less clear: those with higher skill-levels or open sports had a higher accuracy rate than those with lower skill-levels or closed sports.ConclusionEVI involves the visualization of others and the environment, and would be relevant to higher skill-level athletes who engage in open sports. IVI, in contrast, tends to be more self-oriented and would be relevant for utilization by higher skill-level athletes regardless of sport type.
Catalytic bicyclization cascades of oxygen-tethered 1,7-enynes with simple cycloalkanes or aryl sulfonhydrazides have been established via subsequent multiple C-C bond-forming events from alkynyl/alkenyl functions, delivering a series of densely functionalized tetracyclic chromen-2-ones in a functional-group-compatible manner. In the former, Fe-catalyzed spiro-bicyclization involves radical addition, 6-exo-dig cyclization, H-abstraction and 5-endo-trig cyclization sequences under mild conditions, resulting in new spiro-fused cyclopenta[c]chromen-2-ones via dual α,α-C(sp 3 )-H abstractions. The latter enables in-situ sulfonylation and desulfonylation of oxygen-tethered 1,7-enynes to realize the construction of multiple C-C bonds, thereby leading to the formation of naphtho[2,3-c]chromen-6-ones.Example for the synthesis of 5a To a 10-mL Schlenk tube, 2-(phenylethynyl)phenyl methacrylate (1a, 0.20 mmol), tosylhydrazide (4a, 0.7 mmol), TBAI (0.02 mmol), and DCE (3.0 mL) as well as TBHP (5.5 mol/L in decane, 1.0 mmol) were successively added under argon conditions. The reaction system was then stirred at 80 ℃ for 24 h as monitored by TLC. After the completion of the reaction, the resulting mixture was concentrated under vacuum. Purification of the crude prod-Radical-Enabled Bicyclization Cascades of Oxygen-Tethered 1,7-Enynes Chin.
Recently, end-to-end (E2E) models become a competitive alternative to the conventional hybrid automatic speech recognition (ASR) systems. However, they still suffer from speaker mismatch in training and testing condition. In this paper, we use Speech-Transformer (ST) as the study platform to investigate speaker aware training of E2E models. We propose a model called Speaker-Aware Speech-Transformer (SAST), which is a standard ST equipped with a speaker attention module (SAM). The SAM has a static speaker knowledge block (SKB) that is made of i-vectors. At each time step, the encoder output attends to the i-vectors in the block, and generates a weighted combined speaker embedding vector, which helps the model to normalize the speaker variations. The SAST model trained in this way becomes independent of specific training speakers and thus generalizes better to unseen testing speakers. We investigate different factors of SAM. Experimental results on the AISHELL-1 task show that SAST achieves a relative 6.5% CER reduction (CERR) over the speaker-independent (SI) baseline. Moreover, we demonstrate that SAST still works quite well even if the i-vectors in SKB all come from a different data source other than the acoustic training set.
Minimising droplet impact contact time is critical for applications such as self-cleaning, antierosion or anti-icing. Recent studies have used texturing of surfaces to split droplets during impact or inducing asymmetric spreading, but these require specifically designed substrates which cannot be easily reconfigured. A key challenge is to realise an effective reduction in contact time during droplet impingement on a smooth surface without texturing but with an active and programmable control. Our experimental results show that surface acoustic waves (SAWs), generated at a location distant from a point of droplet impact, can be used to minimise contact time by as much as 35% without requiring a textured surface. Besides, the ability to switch on and off the SAWs means that reduction in droplet impact contact time on a surface can be controlled in a programmable manner. Moreover, our results show that by applying acoustic waves, the impact regime of the droplet on the solid surface can be changed from deposition or partial rebound to complete rebound. To study the dynamics of the droplet impact, we developed a numerical model for the multi-phase flow and simulated different droplet impingement scenarios. Numerical results revealed that the acoustic waves could be used to modify and control the internal velocity fields inside the droplet. By breaking the symmetry of the internal recirculation patterns inside the droplet, the kinetic energy recovered from interfacial energy during the retraction process is increased, and the droplet can be fully separated from the surface with a much shorter contact time. Our work opens up opportunities to use SAW devices to minimise the contact time, change the droplet impact regime and program/control the droplet's rebounding on smooth/planar and curved surfaces as well as rough/textured surfaces.
The use of future contextual information is typically shown to be helpful for acoustic modeling. However, for the recurrent neural network (RNN), it's not so easy to model the future temporal context effectively, meanwhile keep lower model latency.In this paper, we attempt to design a RNN acoustic model that being capable of utilizing the future context effectively and directly, with the model latency and computation cost as low as possible. The proposed model is based on the minimal gated recurrent unit (mGRU) with an input projection layer inserted in it. Two context modules, temporal encoding and temporal convolution, are specifically designed for this architecture to model the future context. Experimental results on the Switchboard task and an internal Mandarin ASR task show that, the proposed model performs much better than long short-term memory (LSTM) and mGRU models, whereas enables online decoding with a maximum latency of 170 ms. This model even outperforms a very strong baseline, TDNN-LSTM, with smaller model latency and almost half less parameters.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.