Semi-Supervised Cross-Subject Emotion Recognition Based on Stacked Denoising Autoencoder Architecture Using a Fusion of Multi-Modal Physiological Signals

In recent decades, emotion recognition has received considerable attention. As more enthusiasm has shifted to the physiological pattern, a wide range of elaborate physiological emotion data features come up and are combined with various classifying models to detect one’s emotional states. To circumvent the labor of artificially designing features, we propose to acquire affective and robust representations automatically through the Stacked Denoising Autoencoder (SDA) architecture with unsupervised pre-training, followed by supervised fine-tuning. In this paper, we compare the performances of different features and models through three binary classification tasks based on the Valence-Arousal-Dominance (VAD) affection model. Decision fusion and feature fusion of electroencephalogram (EEG) and peripheral signals are performed on hand-engineered features; data-level fusion is performed on deep-learning methods. It turns out that the fusion data perform better than the two modalities. To take advantage of deep-learning algorithms, we augment the original data and feed it directly into our training model. We use two deep architectures and another generative stacked semi-supervised architecture as references for comparison to test the method’s practical effects. The results reveal that our scheme slightly outperforms the other three deep feature extractors and surpasses the state-of-the-art of hand-engineered features.     

Measurement and analysis of alpha particle induced reactions on gold

Stacked foil activation technique and Ge(Li) gamma ray spectroscopy have been used for the measurement of excitation functions of¹⁹⁷Au(α,xn) (x=1−3),¹⁹⁷Au(α,2pn) and¹⁹⁷Au(α,αn) reactions up to 50 MeV. The experimental cross-sections were compared with the predictions of pre-equilibrium hybrid model, as well as with the more recent index model. A general agreement was found in all reactions using initial exciton numbern ₀=4(4p0h) except for¹⁹⁷Au(α,n) reaction, where index model gives fairly good agreement withn ₀=5(5p0h).     

Hybrid model analysis of the excitation function for alpha induced reaction on121Sb and123Sb

Excitation functions for the reactions¹²¹Sb(α, xn)^125−x I,¹²³Sb(α, xn)^127−x I and¹²¹Sb(α, p3n)¹²¹Te were obtained from the measurements of the residual activity of stacked foils of antimony trioxide evaporated on Al backings from threshold to 60 MeV. The excitation functions for the production of¹²¹I,¹²³I,¹²⁴I,¹²⁶I and¹²¹Te are presented. The experimental data are compared with calculations considering equilibrium as well as pre-equilibrium reaction mechanism according to the hybrid model of Blann. The high energy part of the excitation functions are dominated by the pre-equilibrium reaction mechanism. Calculations were done using a priori calculational method of Overlaid Alice Code of Blann. Most of the excitation functions in the energy range mentioned above could very well be fitted with the hybrid model calculation for exciton numbern=4 withn _n=2 andn _p=2. The overall agreement with the theory is good. Certain discrepancies for example¹²¹Sb(α, p3n)¹²¹Te excitation function, indicate that the production mechanism is different from the one presumed for the calculation.     

Measurement and analysis of alpha-induced reactions on Ta,Ag and Co

Excitation functions for the reaction¹⁸¹Ta (α,xn)^185−x Re,^107,109Ag (α, ypxn) and⁵⁹Co (α, ypxn) were obtained from measurements of residual activity of stacked foils from threshold to 60 MeV. The excitation functions for the production of¹⁸¹Re,¹⁸²Re,¹⁸³Re,¹⁸⁴Re,¹⁰⁵Ag,¹¹¹In,⁵⁴Mn,⁵⁶Co,⁵⁸Co, and⁶⁰Co, are being presented. The experimental data are compared with calculations considering equilibrium as well as pre-equilibrium reactions according to the hybrid model of Blann. High energy part of the excitation functions is dominated by the pre-equilibrium reaction mechanism. Calculations were done using a priori calculational method of Overlaid Alice Code of Blann. Most of the excitation functions in the energy range mentioned above could be very well fitted with the hybrid model calculation for exciton numbern=4 withn _p=2 andn _n=2. The overall agreement with theory is good. Certain discrepancies, however, indicate the necessity to revise the hybrid model with respect to emission of complex particles.     

Structure in the condensed state and amphiphilic properties of novel copolymers having alkyl chains and phosphatidylcholine analogous groups

N‐Stearylacrylamide (SAAm), N‐oleylacrylamide (OAAm), and N‐laurylacrylamide (LAAm) were synthesized. They were characterized by ¹H‐NMR, ¹³C‐NMR, FT‐IR, melting point measurements, and elemental analysis. The copolymerizations of SAAm, OAAm, and LAAm with 2‐[(3‐(acrylamido)propyl)dimethylammonio]ethyl 2′‐isopropyl phosphate were carried out, and a series of amphiphilic poly(acrylamide)s (1a,b, 2, and 3a,b) were obtained. These copolymers showed polyelectrolyte behavior in their viscous properties in polar solvents. X‐ray diffraction analysis indicated that the copolymers 1a,b formed similar stacked bilayers with hydrophilic groups and hydrophobic parts. The polymorphic phase transition of these copolymers was also observed by DSC. In addition, the monolayers as well as LB films of these amphiphilic copolymers were prepared on the surface of water and their π–A isotherms were investigated at different temperatures. © 1999 John Wiley & Sons, Inc. J Polym Sci A: Polym Chem 37: 1293–1302, 1999     

基于WRF数值模式的SDAE-SVR风速预测模型研究

风速预测是风力预报中的核心与基础, 采用天气研究和预报(Weather Research and Forecasting, WRF)模式进行风力预报往往存在风速预测误差较大的问题. 为了提高风速预测精度, 提出了一种基于深度学习和支持向量回归(Support Vector Regression, SVR)相结合的风速预测模型. 该模型以WRF模式预报输出的多种气象变量为基础, 结合气象自动观测站传感器的实测风速, 引入堆栈降噪自动编码(Stacked De-noising Auto-Encoder, SDAE)深度网络来学习样本数据中隐含的深度特征, 然后将该深度网络最后一层输出的深度特征置入回归器SVR中, 利用SVR良好的回归预测性能对WRF模式预报的未来1h风速进行预测订正. 结果表明: 所建立的SDAE-SVR风速预测模型具有较高的风速预测精度, 在对典型日的WRF模式预报未来1h风速的预测订正中, 其平均百分比误差与均方根误差仅为8.28％与0.8 066 m·     

Pruned optical banyan networks on vertical stacking scheme for faster connection establishment

In this paper, we address the issue of faster connection establishment in a large vertically stacked optical Banyan (VSOB) network. The best known global routing algorithm, which turns an N × N crosstalk-free VSOB network into a rearrangeably non-blocking one, has time complexity O (NlogN). This is quite large compared to O (NlogN) time complexity of a single plane banyan network, which is a self-routing network with very high blocking probability. For a large size of switching network this O (NlogN) time complexity may result unacceptably long delay. Therefore, an optical network with very low blocking probability and O (NlogN) time complexity will be useful. Previously proposed Plane Fixed Routing (PFR) algorithm has O (logN) time complexity but results in higher than 2% blocking probability with zero-crosstalk constraint for a network as large as 4096 × 4096 at full load. In this paper, first we propose the pruning of VSOB networks that reduces the hardware cost by almost 30%. The networks can still use the PFR algorithm and results in the same blocking probability. However, we show that the blocking probability can be reduced dramatically while keeping the optimum time complexity O (logN) by allowing only a small amount of crosstalk. Then, we propose a new kind of switching networks in which extra regular banyan planes have been added with the pruned VBOS (P-VSOB) networks. Necessary routing algorithms, namely, PFR_RS and PFR_LS show that this new switching network can reduce the blocking probability to very low value even with zero-crosstalk constraint while keeping the hardware cost 3almost the same as for P-VSOB networks. Both these algorithms also have time complexity O (NlogN).     

基于垂直层叠结构的可见/近红外双波段传感器研究

提出了基于垂直层叠结构的双波段传感器,该结构为同时对可见波段和近红外波段进行成像提供了可能。它的基本原理是利用不同波长的光在硅材料中穿透深度的非线性分布,即:短波长的可见光主要在表面被吸收,长波长的近红外光则主要在更深的位置被吸收。通过垂直层叠结构,抽取不同深度的光生载流子,即可以得到相应波段的成像信息。数值仿真分析表明,结构参量为D1=2μm,D2=18μm的结构能在400~1200 nm波长范围得到响应峰值波长为550 nm和1000 nm的最佳可见/近红外响应。