Signal-background discrimination with convolutional neural networks in the PandaX-III experiment using MC simulation

The PandaX-III experiment will search for neutrinoless double beta decay of ¹³⁶Xe with high pressure gaseous time projection chambers at the China Jin-Ping underground Laboratory. The tracking feature of gaseous detectors helps suppress the background level, resulting in the improvement of the detection sensitivity. We study a method based on the convolutional neural networks to discriminate double beta decay signals against the background f r om high energy gammas generated by ²¹⁴Bi and ²⁰⁸Tl decays based on detailed Monte Carlo simulation. Using the 2-dimensional projections of recorded tracks on two planes, the method successfully suppresses the background level by a factor larger than 100 with a high signal efficiency. An improvement of 62% on the efficiency ratio of \(\in_s/\;\sqrt { \in b} \) is achieved in comparison with the baseline in the PandaX-III conceptual design report.     

Absolute intensity discrimination using two responses

The observer was asked to judge whether a comparison sound was more or less intense than a standard sound of 60 dB SPL. The sounds were broadband noises of 100-ms duration. The independent variables were the range of intensities used in the set of comparison sounds and the presence or absence of feedback. The main dependent variable was the variability of such judgments, which was measured from the psychometric function. The variability, measured in terms of the Weber function, increased from about 1.5 to 4 dB when the stimulus range changed from 10 to 60 dB. Increases in stimulus range increased the variability of the binary judgments in this task, as they had in previous experiments where multiple responses were required. Feedback improved performance primarily at the largest ranges. Only a small fraction of these changes in variability stems from sequential effects.     

Phoneme boundaries in normal and hearing-impaired ears

Two experiments are reported which explore variables that may complicate the interpretation of phoneme boundary data from hearing-impaired listeners. Fourteen synthetic consonant-vowel syllables comprising a/ba-da-ga/ continuum were used as stimuli. The first experiment examined the influence of presentation level and ear of presentation in normal-hearing subjects. Only small differences in the phoneme boundaries and labeling functions were observed between ears and across presentation levels. Thus monaural presentation and relatively high signal level do not appear to be complicating factors in research with hearing-impaired listeners, at least for these stimuli. The second experiment described a test procedure for obtaining phoneme boundaries in some hearing-impaired listeners that controlled for between-subject sources of variation unrelated to hearing impairment and delineated the effects of spectral shaping imposed by the hearing impairment on the labeling functions. Labeling data were obtained from unilaterally hearing-impaired listeners under three test conditions: in the normal ear without any signal distortion; in the normal ear listening through a spectrum shaper that was set to match the subject's suprathreshold audiometric configuration; and in the impaired ear. The reduction in the audibility of the distinctive acoustic/phonetic cues seemed to explain all or part of the effects of the hearing impairment on the labeling functions of some subjects. For many other subjects, however, other forms of distortion in addition to reduced audibility seemed to affect their labeling behavior.     

连续音素的改进深信度网络的识别算法*

为提高连续语音识别中的音素识别率,提出一种基于改进并行回火训练的受限波尔兹曼机的音素识别算法。首先,利用经过等能量划分后的改进并行回火算法来训练受限玻尔兹曼机,接着将受限玻尔兹曼机堆叠组成一个深信度网络,从而作为深度神经网络预训练的基础模型,然后通过softmax层输出,得到用于音素状态后验概率检测的深度神经网络。接着,利用少量的标签数据,根据反向传播算法对网络权重进行微调。最后,将所得后验概率作为隐马尔科夫的发射概率,然后利用Viterbi解码器实现音素识别。在TIMIT语料库上的实验表明,识别率相比于传统的对比散度类算法提高了约4.5%,在不增加计算量的情况下比原始并行回火算法提高约1%。     

Exo-atmospheric target discrimination using probabilistic neural network

Exo-atmospheric targets are especially difficult to distinguish using currently available techniques, because all target parts follow the same spatial trajectory. The feasibility of distinguishing multiple type components of exo-atmospheric targets is demonstrated by applying the probabilistic neural network. Differences in thermal behavior and time-varying signals of space-objects are analyzed during the selection of features used as inputs of the neural network. A novel multi-colorimetric technology is introduced to measure precisely the temporal evolutional characteristics of temperature and emissivity-area products. To test the effectiveness of the recognition algorithm, the results obtained from a set of synthetic multispectral data set are presented and discussed. These results indicate that the discrimination algorithm can obtain a remarkable success rate.     

Multiclass discrimination of cervical precancers using Raman spectroscopy

Raman spectroscopy has the potential to differentiate among the various stages leading to high‐grade cervical cancer such as normal, squamous metaplasia, and low‐grade cancer. For Raman spectroscopy to successfully differentiate among the stages, an applicable statistical method must be developed. Algorithms like linear discriminant analysis (LDA) are incapable of differentiating among three or more types of tissues. We developed a novel statistical method combining the method of maximum representation and discrimination feature (MRDF) to extract diagnostic information with sparse multinomial logistic regression (SMLR) to classify spectra based on nonlinear features for multiclass analysis of Raman spectra. We found that high‐grade spectra classified correctly 95% of the time; low‐grade data classified correctly 74% of the time, improving sensitivity from 92 to 98% and specificity from 81 to 96% suggesting that MRDF with SMLR is a more appropriate technique for categorizing Raman spectra. SMLR also outputs a posterior probability to evaluate the algorithm's accuracy. This combined method holds promise to diagnose subtle changes leading to cervical cancer. Copyright © 2008 John Wiley & Sons, Ltd.     

Phoneme representation and classification in primary auditory cortex

A controversial issue in neurolinguistics is whether basic neural auditory representations found in many animals can account for human perception of speech. This question was addressed by examining how a population of neurons in the primary auditory cortex (A1) of the naive awake ferret encodes phonemes and whether this representation could account for the human ability to discriminate them. When neural responses were characterized and ordered by spectral tuning and dynamics, perceptually significant features including formant patterns in vowels and place and manner of articulation in consonants, were readily visualized by activity in distinct neural subpopulations. Furthermore, these responses faithfully encoded the similarity between the acoustic features of these phonemes. A simple classifier trained on the neural representation was able to simulate human phoneme confusion when tested with novel exemplars. These results suggest that A1 responses are sufficiently rich to encode and discriminate phoneme classes and that humans and animals may build upon the same general acoustic representations to learn boundaries for categorical and robust sound classification.     

Gradient-based PIV using neural networks

This paper proposes a new gradient-based PIV using an artificial neural network for acquiring the characteristics of a two-dimensional flow field. The neural network can effectively realize an accurate approximation of the vector field by introducing some knowledge on the characteristic property. The neural network is trained by using spatial and temporal image gradients so that the basic equation of the gradient-based method is satisfied. Since the neural network itself learns the stream function, the continuity equation of flow is consequently satisfied in the measured velocity vector field. The new gradient-based PIV can be applied to even partly lacking visualized images.     

A packet routing strategy using neural networks on scale-free networks

We propose a dynamic packet routing strategy by using neural networks on scale-free networks. In this strategy, in order to determine the nodes to which the packets should be transmitted, we use path lengths to the destinations of the packets, and adjust the connection weights of the neural networks attached to the nodes from local information and the path lengths. The performances of this strategy on scale-free networks which have the same degree distribution and different degree correlations are compared to one another. Our numerical simulations confirm that this routing strategy is more effective than the shortest path based strategy on scale-free networks with any degree correlations and that the performance of our strategy on assortative scale-free networks is better than that on disassortative and uncorrelated scale-free networks.     

利用波导不变量的浅海负跃层声源深度判别

在夏季负跃层水文条件下,声源激发的声场同时存在折射类和反射类简正波,导致通过分离简正波进行深度判别的方法性能下降甚至失效·提出了一种利用波导不变量判别声源深度的方法。当接收器位于负跃层以下时,水面声源和水下声源的波导不变量数值不同。在本文的仿真条件下,水面声源波导不变量约等于1,当声源位于负跃层以下时,波导不变量大于1.利用频域warping变换从干涉结构中提取与实际距离成线性关系的时域脉冲,可以在未知声源距离的情况下,从声场干涉结构中获得声源激发声场的波导不变量,从而在夏季负跃层水文条件下进行声源深度判别。方法无须声源与接收器存在最近经过距离,可适用于单水听器或水平阵波束形成后输出的LOFAR谱.仿真结果证明,方法可以提取水面目标和水下目标的时延轨迹并计算干涉条纹的波导不变量。最后,对海试实验数据进行分析处理,从水面声源的干涉条纹中提取波导不变量,证明了方法的有效性。     

Analyzing neuronal networks using discrete-time dynamics

We develop mathematical techniques for analyzing detailed Hodgkin-Huxley like models for excitatory-inhibitory neuronal networks. Our strategy for studying a given network is to first reduce it to a discrete-time dynamical system. The discrete model is considerably easier to analyze, both mathematically and computationally, and parameters in the discrete model correspond directly to parameters in the original system of differential equations. While these networks arise in many important applications, a primary focus of this paper is to better understand mechanisms that underlie temporally dynamic responses in early processing of olfactory sensory information. The models presented here exhibit several properties that have been described for olfactory codes in an insect’s Antennal Lobe. These include transient patterns of synchronization and decorrelation of sensory inputs. By reducing the model to a discrete system, we are able to systematically study how properties of the dynamics, including the complex structure of the transients and attractors, depend on factors related to connectivity and the intrinsic and synaptic properties of cells within the network.     

Analysis of pipeline networks using two-ports

Majority of vibration problems arise in pipeline networks are attributed to the high-pressure pulsations. Pulsations are generated by fluid machines such as compressors and pumps. These pulsations turn into shaking forces at elements such as pipe bends and pipe reducers, which in turn excite vibrations in the connected piping network. High vibrations beyond the endurance limit of the pipe material may cause damage to pipes, supports, and equipment. In addition, if the source pulsation frequency coincides with one of the natural frequencies of the piping network, resonance will take place and the vibrations will be magnified to a large scale. Obviously, if these vibrations are not well controlled, they might cause damage to the whole system and foundation, and might lead to substantial financial losses. Thus, prediction of pulsations is important for safe and proper operation. In this paper, a pilot plant equipped with a reciprocating compressor, pipes, bends, and terminated by a vessel is built. The network is modeled using the two-port theory that splits the network into several cascaded elements, and predicts the response of the network. The prediction model uses the measured compressor source data as an input, which is determined by the indirect multi-load method that is usually used to characterize internal combustion engines. A pulsation suppression device is designed, modeled, manufactured and inserted into the pilot network. The pressure pulsations are measured with and without the pulsation suppression device, and compared to the predictions using the two-port theory.     

Reverse-engineering bar charts using neural networks

Journal of Visualization - Reverse-engineering bar charts extract textual and numeric information from the visual representations of bar charts to support application scenarios that require the...     

Modeling plasma equipment using neural networks

Equipment plasma has been modeled semi-empirically using neural networks in conjunction with statistical experimental design. A 3³ factorial design was employed to characterize the plasma, in which the variables that were varied include a source power, pressure, and Ar flow rate. As a test data for model validation, 16 experiments were additionally conducted. A total of six plasma attributes were modeled, which include electron density, electron temperature, and plasma potential as well as their spatial uniformities. A planar, inductively coupled plasma was generated in a multipole plasma etch equipment and Langmuir probe was utilized for data collection. Root mean-squared prediction errors measured on the test data are 0.323 (10 ¹¹/cm³), 0.267 (eV) and 1.141 (V) for electron density, electron temperature, and plasma potential, respectively. Comparisons with a statistical response surface model (RSM) revealed that neural network models are more accurate by an improvement of more than 25% in prediction performance. A similar level of prediction accuracy was also achieved in modeling spatial uniformity data. Consequently, neural networks demonstrated much better prediction capabilities over RSM in modeling complex equipment plasma     

Demonstrating superior discrimination of locally prepared states using nonlocal measurements

We experimentally demonstrate the superior discrimination of separated, unentangled two-qubit correlated states using nonlocal measurements, when compared with measurements based on local operations and classical communications. When predicted theoretically, this phenomenon was dubbed "quantum nonlocality without entanglement." We characterize the performance of the nonlocal, or joint, measurement with a payoff function, for which we measure 0.72 +/- 0.02, compared with the maximum locally achievable value of 2/3 and the overall optimal value of 0.75.     

Dark matter experiment using Argon pulse-shape discrimination: DEAP-3600

DEAP-3600 is a single-phase liquid-argon dark matter detector that at time of writing is cooling down in preparation for filling at the SNOLAB facility near Sudbury, Ontario, Canada. DEAP-3600 is designed and constructed to achieve a sensitivity of 10^?46cm² for a WIMP-nucleon cross section for a 100 GeV WIMP. The steps taken in design and construction to achieve the ultra-low backgrounds required for such a sensitive WIMP search are reviewed.     

Optimal unambiguous discrimination of pure quantum states using SDP method

In the present paper, an exact analytic solution for the optimal unambiguous state discrimination(OPUSD) problem involving an arbitrary number of pure linearly independent quantum states with real and complex inner product is presented. Using semidefinite programming and Karush-Kuhn-Tucker convex optimization method, we derive an analytical formula which shows the relation between optimal solution of unambiguous state discrimination problem and an arbitrary number of pure linearly independent quantum states.