Qisong Wang

Unsupervised domain adaptation techniques based on auto-encoder for non-stationary EEG-based emotion recognition

In electroencephalography (EEG)-based emotion recognition systems, the distribution between the training samples and the testing samples may be mismatched if they are sampled from different experimental sessions or subjects because of user fatigue, different electrode placements, varying impedances, etc. Therefore, it is difficult to directly classify the EEG patterns with a conventional classifier. The domain adaptation method, which is aimed at obtaining a common representation across training and test domains, is an effective method for reducing the distribution discrepancy. However, the existing domain adaptation strategies either employ a linear transformation or learn the nonlinearity mapping without a consistency constraint; they are not sufficiently powerful to obtain a similar distribution from highly non-stationary EEG signals. To address this problem, in this paper, a novel component, called the subspace alignment auto-encoder (SAAE), is proposed. Taking advantage of both nonlinear transformation and a consistency constraint, we combine an auto-encoder network and a subspace alignment solution in a unified framework. As a result, the source domain can be aligned with the target domain together with its class label, and any supervised method can be applied to the new source domain to train a classifier for classification in the target domain, as the aligned source domain follows a distribution similar to that of the target domain. We compared our SAAE method with six typical approaches using a public EEG dataset containing three affective states: positive, neutral, and negative. Subject-to-subject and session-to-session evaluations were performed. The subject-to-subject experimental results demonstrate that our component achieves a mean accuracy of 77.88% in comparison with a state-of-the-art method, TCA, which achieves 73.82% on average. In addition, the average classification accuracy of SAAE in the session-to-session evaluation for all the 15 subjects in a dataset is 81.81%, an improvement of up to 1.62% on average as compared to the best baseline TCA. The experimental results show the effectiveness of the proposed method relative to state-of-the-art methods. It can be concluded that SAAE is a useful and effective tool for decreasing domain discrepancy and reducing performance degradation across subjects and sessions in the EEG-based emotion recognition field.

A Fast, Efficient Domain Adaptation Technique for Cross-Domain Electroencephalography(EEG)-Based Emotion Recognition

Electroencephalography (EEG)-based emotion recognition is an important element in psychiatric health diagnosis for patients. However, the underlying EEG sensor signals are always non-stationary if they are sampled from different experimental sessions or subjects. This results in the deterioration of the classification performance. Domain adaptation methods offer an effective way to reduce the discrepancy of marginal distribution. However, for EEG sensor signals, both marginal and conditional distributions may be mismatched. In addition, the existing domain adaptation strategies always require a high level of additional computation. To address this problem, a novel strategy named adaptive subspace feature matching (ASFM) is proposed in this paper in order to integrate both the marginal and conditional distributions within a unified framework (without any labeled samples from target subjects). Specifically, we develop a linear transformation function which matches the marginal distributions of the source and target subspaces without a regularization term. This significantly decreases the time complexity of our domain adaptation procedure. As a result, both marginal and conditional distribution discrepancies between the source domain and unlabeled target domain can be reduced, and logistic regression (LR) can be applied to the new source domain in order to train a classifier for use in the target domain, since the aligned source domain follows a distribution which is similar to that of the target domain. We compare our ASFM method with six typical approaches using a public EEG dataset with three affective states: positive, neutral, and negative. Both offline and online evaluations were performed. The subject-to-subject offline experimental results demonstrate that our component achieves a mean accuracy and standard deviation of 80.46% and 6.84%, respectively, as compared with a state-of-the-art method, the subspace alignment auto-encoder (SAAE), which achieves values of 77.88% and 7.33% on average, respectively. For the online analysis, the average classification accuracy and standard deviation of ASFM in the subject-to-subject evaluation for all the 15 subjects in a dataset was 75.11% and 7.65%, respectively, gaining a significant performance improvement compared to the best baseline LR which achieves 56.38% and 7.48%, respectively. The experimental results confirm the effectiveness of the proposed method relative to state-of-the-art methods. Moreover, computational efficiency of the proposed ASFM method is much better than standard domain adaptation; if the numbers of training samples and test samples are controlled within certain range, it is suitable for real-time classification. It can be concluded that ASFM is a useful and effective tool for decreasing domain discrepancy and reducing performance degradation across subjects and sessions in the field of EEG-based emotion recognition.

A multifunctional wireless body area sensors network with real time embedded data analysis

Quantitative limitation of sensor nodes, single function and limited data processing capability, are the major problems of traditional Wireless Body Area Network (WBAN). In many applications, numbers of sensor nodes used for multi-parameter synchronous measurements are necessary, which usually require real-time analysis and high portability. This paper presents a multifunctional WBAN with real time embedded data analysis ability. Each sensor node in the WBAN is capable of EEG/ECG/EOG, respiration impedance and body posture measurement. The DSP-embedded node, namely the main node, is solely responsible for the on-line data processing. The experiments demonstrated that this platform is able to afford more than 16 nodes simultaneously, which maintains a low data loss rate and high signal quality even for high sampling rate applications. With the advantage of DSP core, this platform performs perfectly in processing complex data acquired by multiple sensor nodes in real time.

Optimal detection strategy for super-resolving quantum lidar

The description of quantum lidar in the presence of photon loss and phase noise is presented. Taylor series is directly exploited to expand the interference signal to separate the detected phase and the phase noise. The analytical expression of interference signal and its sensitivity are illustrated by binary outcome homodyne, parity photon counting, and zero-nonzero photon counting detection. Numerical calculation indicates that homodyne detection has the best sensitivity and resolution and should be considered as the optimal detection strategy for quantum lidar in the diffusion region of κ<10−2. However, parity detection should be the best detection scheme for resolution, and zero-nonzero detection represents the optimal detection for sensitivity in the rest region. Finally, zero-nonzero detection produces better sensitivity than parity detection.

Least squares support vector machine based lithium battery capacity prediction

The capacity character of lithium-ion battery is one of the most important performance parameters, which need to be accurate measurement for the safety and efficiency usage. In this paper, the regularity that battery capacity parameter changes with working temperature and charge or discharge rate has been analyzed, and the least squares support vector machine based battery capacity prediction method has been proposed for LiFePO4 battery. Furthermore, the lithium-ion battery capacity estimation experiments are carried out for both charging and discharging process, and the related results illustrate that the proposed method is able to give an accurate and efficient estimation of the corresponding battery capacity parameter in the presumed range of working temperature and charge or discharge rate.

Thin Plate Spline-Based Coulombic Efficiency Prediction of Lithium Battery

The performance of lithium-ion rechargeable battery depends on its coulombic efficiency, which determines the measurement accuracy of the state of charge as well. In this context, the regularity that coulombic efficiency changes with environmental temperature and charge-discharge rate was analyzed in different cases, and a TPS (Thin Plate Spline) surface fitting-based coulombic efficiency estimation method was proposed for lithium-ion rechargeable battery. The presented method has a unique advantage in 3D smooth surface interpolation, which made it possible to model the functional relationship between battery coulombic efficiency, ambient temperature and charge-discharge rate, and then adjust the reconstructed precision with the plate rigidity sequentially. The experimental results illustrate that the proposed method is able to estimate the corresponding coulombic efficiency in the presumed range of ambient temperature and charge-discharge rate without battery cycling procedure. Furthermore, the accuracy and promptness of the novel method are verified simultaneously.

Imitation-tumor targeting based on continuous-wave near-infrared tomography

Abstract Continuous-wave Near-Infrared (NIR) optical spectroscopy has shown great diagnostic capability in the early tumor detection with advantages of low-cost, portable, non-invasive, and non-radiative. In this paper, Modified Lambert-Beer Theory is deployed to address the low-resolution issues of the NIR technique and to design the tumor detecting and imaging system. Considering that tumor tissues have features such as high blood flow and hypoxia, the proposed technique can detect the location, size, and other information of the tumor tissues by comparing the absorbance between pathological and normal tissues. Finally, the tumor tissues can be imaged through tomographic method. The simulation experiments prove that the proposed technique and designed system can efficiently detect the tumor tissues, achieving imaging precision within 1 mm. The work of the paper has shown great potential in the diagnosis of tumor close to body surface.

Tissue phantom-based breast cancer detection using continuous near-infrared sensor

ABSTRACT Womens health is seriously threatened by breast cancer. Taking advantage of efficient diagnostic instruments to identify the disease is very meaningful in prolonging life. As a cheap noninvasive radiation-free technology, Near-infrared Spectroscopy is suitable for general breast cancer examination. A discrimination method of breast cancer is presented using the deference between absorption coefficients and applied to construct a blood oxygen detection device based on Modified Lambert-Beer theory. Combined with multi-wavelength multi-path near-infrared sensing technology, the proposed method can quantitatively distinguish the normal breast from the abnormal one by measuring the absorption coefficients of breast tissue and the blood oxygen saturation. An objective judgment about the breast tumor is made according to its high absorption of near-infrared light. The phantom experiment is implemented to show the presented method is able to recognize the absorption differences between phantoms and demonstrates its feasibility in the breast tumor detection.

An effective feature extraction method by power spectral density of EEG signal for 2-class motor imagery-based BCI

EEG signals have weak intensity, low signal-to-noise ratio, non-stationary, non-linear, time-frequency-spatial characteristics. Therefore, it is important to extract adaptive and robust features that reflect time, frequency and spatial characteristics. This paper proposes an effective feature extraction method WDPSD (feature extraction from the Weighted Difference of Power Spectral Density in an optimal channel couple) that can reflect time, frequency and spatial characteristics for 2-class motor imagery-based BCI system. In the WDPSD method, firstly, Power Spectral Density (PSD) matrices of EEG signals are calculated in all channels, and an optimal channel couple is selected from all possible channel couples by checking non-stationary and class separability, and then a weight matrix which reflects non-stationary of PSD difference matrix in selected channel couple is calculated; finally, the robust and adaptive features are extracted from the PSD difference matrix weighted by the weight matrix. The proposed method is evaluated from EEG signals of BCI Competition IV Dataset 2a and Dataset 2b. The experimental results show a good classification accuracy in single session, session-to-session, and the different types of 2-class motor imagery for different subjects.

Evoked Hemodynamic Response Estimation to Auditory Stimulus Using Recursive Least Squares Adaptive Filtering with Multidistance Measurement of Near-Infrared Spectroscopy

The performance of functional near-infrared spectroscopy (fNIRS) is sometimes degraded by the interference caused by the physical or the systemic physiological activities. Several interferences presented during fNIRS recordings are mainly induced by cardiac pulse, breathing, and spontaneous physiological low-frequency oscillations. In previous work, we introduced a multidistance measurement to reduce physiological interference based on recursive least squares (RLS) adaptive filtering. Monte Carlo simulations have been implemented to evaluate the performance of RLS adaptive filtering. However, its suitability and performance on human data still remain to be evaluated. Here, we address the issue of how to detect evoked hemodynamic response to auditory stimulus using RLS adaptive filtering method. A multidistance probe based on continuous wave fNIRS is devised to achieve the fNIRS measurement and further study the brain functional activation. This study verifies our previous findings that RLS adaptive filtering is an effective method to suppress global interference and also provides a practical way for real-time detecting brain activity based on multidistance measurement.