• Journal of Intelligence and Information Systems
• /
• v.25 no.1
• /
• pp.163-177
• /
• 2019

As smartphones are getting widely used, human activity recognition (HAR) tasks for recognizing personal activities of smartphone users with multimodal data have been actively studied recently. The research area is expanding from the recognition of the simple body movement of an individual user to the recognition of low-level behavior and high-level behavior. However, HAR tasks for recognizing interaction behavior with other people, such as whether the user is accompanying or communicating with someone else, have gotten less attention so far. And previous research for recognizing interaction behavior has usually depended on audio, Bluetooth, and Wi-Fi sensors, which are vulnerable to privacy issues and require much time to collect enough data. Whereas physical sensors including accelerometer, magnetic field and gyroscope sensors are less vulnerable to privacy issues and can collect a large amount of data within a short time. In this paper, a method for detecting accompanying status based on deep learning model by only using multimodal physical sensor data, such as an accelerometer, magnetic field and gyroscope, was proposed. The accompanying status was defined as a redefinition of a part of the user interaction behavior, including whether the user is accompanying with an acquaintance at a close distance and the user is actively communicating with the acquaintance. A framework based on convolutional neural networks (CNN) and long short-term memory (LSTM) recurrent networks for classifying accompanying and conversation was proposed. First, a data preprocessing method which consists of time synchronization of multimodal data from different physical sensors, data normalization and sequence data generation was introduced. We applied the nearest interpolation to synchronize the time of collected data from different sensors. Normalization was performed for each x, y, z axis value of the sensor data, and the sequence data was generated according to the sliding window method. Then, the sequence data became the input for CNN, where feature maps representing local dependencies of the original sequence are extracted. The CNN consisted of 3 convolutional layers and did not have a pooling layer to maintain the temporal information of the sequence data. Next, LSTM recurrent networks received the feature maps, learned long-term dependencies from them and extracted features. The LSTM recurrent networks consisted of two layers, each with 128 cells. Finally, the extracted features were used for classification by softmax classifier. The loss function of the model was cross entropy function and the weights of the model were randomly initialized on a normal distribution with an average of 0 and a standard deviation of 0.1. The model was trained using adaptive moment estimation (ADAM) optimization algorithm and the mini batch size was set to 128. We applied dropout to input values of the LSTM recurrent networks to prevent overfitting. The initial learning rate was set to 0.001, and it decreased exponentially by 0.99 at the end of each epoch training. An Android smartphone application was developed and released to collect data. We collected smartphone data for a total of 18 subjects. Using the data, the model classified accompanying and conversation by 98.74% and 98.83% accuracy each. Both the F1 score and accuracy of the model were higher than the F1 score and accuracy of the majority vote classifier, support vector machine, and deep recurrent neural network. In the future research, we will focus on more rigorous multimodal sensor data synchronization methods that minimize the time stamp differences. In addition, we will further study transfer learning method that enables transfer of trained models tailored to the training data to the evaluation data that follows a different distribution. It is expected that a model capable of exhibiting robust recognition performance against changes in data that is not considered in the model learning stage will be obtained.

• The Korean Journal of Nuclear Medicine
• /
• v.39 no.6
• /
• pp.413-420
• /
• 2005

Purpose: Neuroreceptor PET studies require 60-120 minutes to complete and head motion of the subject during the PET scan increases the uncertainty in measured activity. In this study, we investigated the effects of the data-driven head mutton correction on the evaluation of endogenous dopamine release (DAR) in the striatum during the motor task which might have caused significant head motion artifact. Materials and Methods: $[^{11}C]raclopride$ PET scans on 4 normal volunteers acquired with bolus plus constant infusion protocol were retrospectively analyzed. Following the 50 min resting period, the participants played a video game with a monetary reward for 40 min. Dynamic frames acquired during the equilibrium condition (pre-task: 30-50 min, task: 70-90 min, post-task: 110-120 min) were realigned to the first frame in pre-task condition. Intra-condition registrations between the frames were performed, and average image for each condition was created and registered to the pre-task image (inter-condition registration). Pre-task PET image was then co-registered to own MRI of each participant and transformation parameters were reapplied to the others. Volumes of interest (VOI) for dorsal putamen (PU) and caudate (CA), ventral striatum (VS), and cerebellum were defined on the MRI. Binding potential (BP) was measured and DAR was calculated as the percent change of BP during and after the task. SPM analyses on the BP parametric images were also performed to explore the regional difference in the effects of head motion on BP and DAR estimation. Results: Changes in position and orientation of the striatum during the PET scans were observed before the head motion correction. BP values at pre-task condition were not changed significantly after the intra-condition registration. However, the BP values during and after the task and DAR were significantly changed after the correction. SPM analysis also showed that the extent and significance of the BP differences were significantly changed by the head motion correction and such changes were prominent in periphery of the striatum. Conclusion: The results suggest that misalignment of MRI-based VOI and the striatum in PET images and incorrect DAR estimation due to the head motion during the PET activation study were significant, but could be remedied by the data-driven head motion correction.