Media Research Center

About: Media Research Center is a based out in . It is known for research contribution in the topics: Collaborative learning & Educational technology. The organization has 491 authors who have published 950 publications receiving 28581 citations. The organization is also known as: MRC.

Do not cross the line : heuristic spatial updating in dynamic scenes

Abstract: In two experiments, we examined the spatial integration of two viewpoints onto dynamic scenes. We tested the spatial-alignment hypothesis (which predicts integration by alignment along the shortest path) against the spatial-heuristic hypothesis (which predicts integration by observation of the left–right orientation on the screen). The stimuli consisted of film clips comprising two shots, each showing a car driving by. In Experiment 1, the dynamic scenes were ambiguous with regard to their interpretation: The cars could have been driving either in the same or in opposite directions. In line with the spatial-heuristic hypothesis, participants responded with “same direction” if the cars shared a screen direction. In Experiment 2, environmental cues disambiguated the dynamic scenes, and the screen direction of the cars was either congruent or incongruent with the depicted environmental cues. As compared with congruent dynamic scenes, incongruent dynamic scenes elicited prolonged reaction times, thus suggesting that heuristic spatial updating was used with congruent stimuli, whereas spatial-alignment processes were used with incongruent stimuli.

Robust query-by-singing/humming system against background noise environments

Abstract: Under background noise environments, the performance of the Query-by-Singing/Humming (QbSH) system is considerably degraded. Since human pitch information is used as a feature vector for the QbSH system, a noise robust pitch estimation algorithm is inevitable. Thus, a novel pitch-estimation method is proposed by integrating temporal-autocorrelation and spectral-salience methods. As a pre-processing block, spectral smoothing is applied to enhance the stationarity of the noisy input signal. To calculate the similarity between the MIDI database and input humming signal, the dynamic time warping (DTW) algorithm is used. Jang's corpus and AURORA2 database are selected as humming and background noise signals, respectively. Compared with the standard pitch estimation algorithm in the ITU-T G.729 speech codec, the proposed pitch estimation method improves the average accuracy by 11.7% for the 0 dB signal-to-noise ratio (SNR) noise case. It also improves top-20 ratio and mean reciprocal rank (MRR) of the proposed QbSH system, on average, by 7.4% and 0.13, respectively.

Magnitude representation in sequential comparison of two-digit numbers is not holistic either

Abstract: There is accumulating evidence suggesting that two-digit number magnitude is represented in a decomposed fashion into tens and units rather than holistically as one integrated entity. However, recently, it has been claimed that this property does not hold for the case when two to-be-compared numbers are presented sequentially. In the present study, we pursued this issue in two experiments by evaluating perceptual as well as strategic aspects arising for sequential stimulus presentation in a magnitude comparison task. We observed reliable unit-decade compatibility effects indicating decomposed processing of tens and units in a magnitude comparison task with sequential presentation of the to-be-compared numbers. In particular, we found that both confounding low-level perceptual features and stimulus set characteristics determining cue validity of the units influenced the compatibility effect. Taken together, our results clearly indicate that decomposed representations of tens and units seem to be a general characteristic of multi-digit number magnitude processing, rather than an exception occurring under very specific conditions only. Implications of these results for the understanding of number magnitude representations are discussed.

Cloud-Edge Collaboration Framework for IoT data analytics

Abstract: In recent years, big data analysis from Internet of Things (IoT) has been getting more attention. Several cloud platforms have provided machine learning service with a pre-trained model to understand IoT data. However, it is necessary to transfer personal data in order to use the cloud service, and network problems might be preventing the customer from getting analysis results at an appropriate time. To overcome these problems, data and analysis task are moving to the edge platform. However, most edge devices do not have enough capacity to process and train large amounts of data. In this paper, we propose a new framework structure that can analyze IoT data by distributing analysis role. The proposed framework is designed to maximize the resources of the cloud to generate the model and to use the model at the edge to enable immediate and instantaneous actuator operation. And we also present a case study to verify this framework.