Summary
Collaborative filtering (CF) is a widely used technique in recommender systems. With rapid development in deep learning, neural network‐based CF models have gained great attention in the recent years, especially autoencoder‐based CF model. Although autoencoder‐based CF model is faster compared with some existing neural network‐based models (eg, Deep Restricted Boltzmann Machine‐based CF), it is still impractical to handle extremely large‐scale data. In this paper, we practically verify that most non‐zero entries of the input matrix are concentrated in a few rows. Considering this sparse characteristic, we propose a new method for training autoencoder‐based CF. We run experiments on two popular datasets MovieLens 1 M and MovieLens 10 M. Experimental results show that our algorithm leads to orders of magnitude speed‐up for training (stacked) autoencoder‐based CF model while achieving comparable performance compared with existing state‐of‐the‐art models.
To investigate and compare the common features and differences of the cognitive processes, during which interfaces with diverse similarities are evaluated, this article chose the color code and layout forms of digital interface to carry out further research. The study adopted the visual Oddball experimental paradigm that was based on the event-related potential technique and integrated the behavioral and event-related potential data to analyze the neural features of the cognitive process when two coding forms were individually processed. The result reveals that there were P300 components, elicited by the target stimuli, in both of the two experiment sessions. The average amplitude of P300 positively correlates the similarities between the target and standard stimuli, with its latency positively correlating the overall complexity of the stimuli. In the color experiment session, there was apparent visual mismatching negativity around 200 ms after the present of the target, which is related with the early attention. The empirical significance of conclusions drawn in this study is listed as follows: first, it can help to effectively evaluate the usability of guiding features in the digital interfaces through the investigation on visual mismatching negativity elicited in the early attention process; second, the amplitude and latency of the P300 component can be applied in the evaluation and filtering of design schemes, which is based on the similarities perceived in the iterative process and this would enhance efficiency of user interface designers.
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.