LIU Yinuo

Our department attaches great importance to the research training of students, and most courses have projects. Here are some course projects done in the previous year.

Note: These project reports or slides are all in Chinese.

Selected Coursework

Binary Classification of Blurry Images by Using Convolutional Neural Networks

I used AlexNet and VGG-16 to do the binary classification of blurry images. There are two training methods, that is, whether to pre-train on clear images or not. Results show that the pre-trained model can get higher accuracy and better performance. But the visualized feature filters and convolutional layers do not show features similar to the human visual system, due to the small number of training data.

see: report, slides

Indecision Trait Measurement in Short-term Decision-making

Our group tried to make a questionnaire for measuring indecision in short-term decision-making. We improved our draft version through literature review, factor analysis, structural equation modeling, and data simulation.

see: slides

Absolute Sensory Thresholds for Different Emotional Faces

The sensory thresholds of facial stimuli with different emotional valences were measured by psychophysical methods. It was proved that there was a lower sensory threshold for negative emotions, but there were individual differences in the thresholds. By introducing and adopting a subjective rating scale, it was proved that people can process emotional information at the unconscious level.

see: report

Exploring the Effectiveness of Using Different Physiological Indicators to Detect Lies

Using the simulated theft scenarios combined with the GKT polygraph mode, it was verified that the electroskin index has the highest sensitivity in detecting lies. Through support vector machine modeling, it was also found that the model results under different polygraph standards are different, but the skin electrodermal indicators have a stable and high accuracy.

see: report