RESUMO
Recently, we introduced a current limiter-based novel transcranial direct-current stimulation (tDCS) device that does not generate significant tDCS-induced electrical artifacts, thereby facilitating simultaneous electroencephalography (EEG) measurement during tDCS application. In this study, we investigated the neuromodulatory effect of the tDCS device using resting-state EEG data measured during tDCS application in terms of EEG power spectral densities (PSD) and brain network indices (clustering coefficient and path length). Resting-state EEG data were recorded from 10 healthy subjects during both eyes-open (EO) and eyes-closed (EC) states for each of five different conditions (baseline, sham, post-sham, tDCS, and post-tDCS). In the tDCS condition, tDCS was applied for 12 min with a current intensity of 1.5 mA, whereas tDCS was applied only for the first 30 s in the sham condition. EEG PSD and brain network indices were computed for the alpha frequency band most closely associated with resting-state EEG. Both alpha PSD and network indices were found to significantly increase during and after tDCS application compared to those of the baseline condition in the EO state, but not in the EC state owing to the ceiling effect. Our results demonstrate the neuromodulatory effect of the tDCS device that does not generate significant tDCS-induced electrical artifacts, thereby allowing simultaneous measurement of electrical brain activity. We expect our novel tDCS device to be practically useful in exploring the impact of tDCS on neuromodulation more precisely using ongoing EEG data simultaneously measured during tDCS application.
RESUMO
Facial emotion recognition (FER) systems are imperative in recent advanced artificial intelligence (AI) applications to realize better human-computer interactions. Most deep learning-based FER systems have issues with low accuracy and high resource requirements, especially when deployed on edge devices with limited computing resources and memory. To tackle these problems, a lightweight FER system, called Light-FER, is proposed in this paper, which is obtained from the Xception model through model compression. First, pruning is performed during the network training to remove the less important connections within the architecture of Xception. Second, the model is quantized to half-precision format, which could significantly reduce its memory consumption. Third, different deep learning compilers performing several advanced optimization techniques are benchmarked to further accelerate the inference speed of the FER system. Lastly, to experimentally demonstrate the objectives of the proposed system on edge devices, Light-FER is deployed on NVIDIA Jetson Nano.
Assuntos
Reconhecimento Facial , Humanos , Expressão Facial , Emoções , Inteligência ArtificialRESUMO
The rapid development of deep-learning-based edge artificial intelligence applications and their data-driven nature has led to several research issues. One key issue is the collaboration of the edge and cloud to optimize such applications by increasing inference speed and reducing latency. Some researchers have focused on simulations that verify that a collaborative edge-cloud network would be optimal, but the real-world implementation is not considered. Most researchers focus on the accuracy of the detection and recognition algorithm but not the inference speed in actual deployment. Others have implemented such networks with minimal pressure on the cloud node, thus defeating the purpose of an edge-cloud collaboration. In this study, we propose a method to increase inference speed and reduce latency by implementing a real-time face recognition system in which all face detection tasks are handled on the edge device and by forwarding cropped face images that are significantly smaller than the whole video frame, while face recognition tasks are processed at the cloud. In this system, both devices communicate using the TCP/IP protocol of wireless communication. Our experiment is executed using a Jetson Nano GPU board and a PC as the cloud. This framework is studied in terms of the frame-per-second (FPS) rate. We further compare our framework using two scenarios in which face detection and recognition tasks are deployed on the (1) edge and (2) cloud. The experimental results show that combining the edge and cloud is an effective way to accelerate the inferencing process because the maximum FPS achieved by the edge-cloud deployment was 1.91× more than the cloud deployment and 8.5× more than the edge deployment.
Assuntos
Inteligência Artificial , Reconhecimento Facial , Algoritmos , Sistemas ComputacionaisRESUMO
In this paper, we derive the average bit error rate (BER) of subcarrier multiplexing (SCM)-based free space optics (FSO) systems using a dual-drive Mach-Zehnder modulator (DD-MZM) for optical single-sideband (OSSB) signals under atmospheric turbulence channels. In particular, we consider the third-order intermodulation (IM3), a significant performance degradation factor, in the case of high input signal power systems. The derived average BER, as a function of the input signal power and the scintillation index, is employed to determine the optimum number of SCM users upon the designing FSO systems. For instance, when the user number doubles, the input signal power decreases by almost 2 dBm under the log-normal and exponential turbulence channels at a given average BER.
RESUMO
This paper analytically investigates a bit error rate (BER) performance of radio over free space optical (FSO) systems considering laser phase noise under Gamma-Gamma turbulence channels. An external modulation using a dual drive Mach-Zehnder modulator (DD-MZM) and a phase shifter is employed because a DD-MZM is robust against a laser chirp and provides high spectral efficiency. We derive a closed form average BER as a function of different turbulence strengths and laser diode (LD) linewidth, and investigate its analytical behavior under practical scenario. As a result, for a given average SNR with normalized perturbation, it is shown that the difference of average BER corresponding to two LDs (with linewidth of 624 MHz and 10 MHz) under weak turbulence is almost 3 times larger than that under strong turbulence.