Beyond nonutilization: Irrelevant cues can gate learning in probabilistic categorization.

In probabilistic categorization, also known as multiple cue probability learning (MCPL), people learn to predict a discrete outcome on the basis of imperfectly valid cues. In MCPL, normatively irrelevant cues are usually ignored, which stands in apparent conflict with recent research in deterministic categorization that has shown that people sometimes use irrelevant cues to gate access to partial knowledge encapsulated in independent partitions. The authors report 2 experiments that sought support for the existence of such knowledge partitioning in probabilistic categorization. The results indicate that, as in other areas of concept acquisition (such as function learning and deterministic categorization), a significant proportion of participants partitioned their knowledge on the basis of an irrelevant cue. The authors show by computational modeling that knowledge partitioning cannot be accommodated by 2 exemplar models (Generalized Context Model and Rapid Attention Shifts 'N Learning), whereas a rule-based model (General Recognition Theory) can capture partitioned performance. The authors conclude by pointing to the necessity of a mixture-of-experts approach to capture performance in MCPL and by identifying reduction of complexity as a possible explanation for partitioning. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

贝叶斯网络结构学习方法新探

从大型数据库中学习网络结构一直是贝叶斯网络学习的研究热点.针对此问题提出了一种基于预测能力的学习算法,通过预测能力建立并调整贝叶斯网络结构,把变量之间弧的存在性与方向有机地结合在一起。     

Genetic-Fuzzy Approach for Modeling Complex Systems with an Example Application in Masonry Bond Strength Prediction

A genetic-fuzzy learning from examples (GFLFE) approach is presented for determining fuzzy rule bases generated from input/output data sets. The method is less computationally intensive than existing fuzzy rule base learning algorithms as the optimization variables are limited to the membership function widths of a single rule, which is equal to the number of input variables to the fuzzy rule base. This is accomplished by primary width optimization of a fuzzy learning from examples algorithm. The approach is demonstrated by a case study in masonry bond strength prediction. This example is appropriate as theoretical models to predict masonry bond strength are not available. The GFLFE method is compared to a similar learning method using constrained nonlinear optimization. The writers’ results indicate that the use of a genetic optimization strategy as opposed to constrained nonlinear optimization provides significant improvement in the fuzzy rule base as indicated by a reduced fitness (objective) function and reduced root-mean-squared error of an evaluation data set.     

多分辨率小波极限学习机

针对一类具有空间不均匀性的辨识和回归问题,提出了基于小波分析的极限学习机方法.从多分辨率分析的思想出发,构造一簇紧支撑正交小波作为隐层激活函数,并利用改进的误差最小化极限学习机训练输出层权重,避免了新加入高分辨率子网络后的重新训练.同时,由一维多分辨分析的张量积构造了二维多分辨小波极限学习机.进而通过脊波变换将小波学习机扩展到高维空间,对脊波函数的伸缩、方向和位置参数进行优化计算.对具有奇异性的函数仿真结果证明,与标准极限学习机相比,小波极限学习机由于其聚微性能在极短的训练时间内更好地逼近目标.一些实际基准回归问题上的测试验证了脊波极限学习机在其中大部分问题上达到更高的训练和泛化精度.      

Efficient adaptive learning for classification tasks with binary units

This article presents a new incremental learning algorithm for classification tasks, called NetLines, which is well adapted for both binary and real-valued input patterns. It generates small, compact feedforward neural networks with one hidden layer of binary units and binary output units. A convergence theorem ensures that solutions with a finite number of hidden units exist for both binary and real-valued input patterns. An implementation for problems with more than two classes, valid for any binary classifier, is proposed. The generalization error and the size of the resulting networks are compared to the best published results on well-known classification benchmarks. Early stopping is shown to decrease overfitting, without improving the generalization performance.     

The costs of supervised classification: The effect of learning task on conceptual flexibility.

Research has shown that learning a concept via standard supervised classification leads to a focus on diagnostic features, whereas learning by inferring missing features promotes the acquisition of within-category information. Accordingly, we predicted that classification learning would produce a deficit in people's ability to draw novel contrasts—distinctions that were not part of training—compared with feature inference learning. Two experiments confirmed that classification learners were at a disadvantage at making novel distinctions. Eye movement data indicated that this conceptual inflexibility was due to (a) a narrower attention profile that reduces the encoding of many category features and (b) learned inattention that inhibits the reallocation of attention to newly relevant information. Implications of these costs of supervised classification learning for views of conceptual structure are discussed. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Ductility prediction of Ti aluminide intermetallics through neuro-fuzzy set approach

Several studies have been conducted to improve the room temperature ductility of titanium aluminide intermetallics through alloy design and microstructure modifications. Ductility of two phase (α₂+γ) binary Ti aluminide intermetallics centered on Ti-48Al (at%) was reported as maximum (∼1.5%) in desirable heat treatment condition and so more studies were attempted near to this composition. In the present work also, ductility has been studied for the alloy variants of this composition through fuzzy modeling. Neuro-fuzzy models were developed through Adaptive Neural Fuzzy Inference System (ANFIS) using subtractive cluster techniques. The input parameters were fuzzified with Gaussian membership functions to develop the fuzzy rules. The output of each rule was obtained by the evaluation of the membership values. Finally the overall fuzzy model response was obtained as the weighted average of the individual rule response. Ductility database were prepared and three parameters viz. alloy type, grain size and heat treatment cycle were selected for modeling. Additional, ductility data were generated from literature based experimental data for training and validation of models on the basis of linearity and considering the primary effect of these three parameters. Adequacy of developed models was evaluated with the generated data sets. Different evaluation measures were considered and the resulting graphs from the developed model were analyzed. The results of the fuzzy models were found to be very close to the literature based generated data and it also showed the possibility of improving ductility upto 7% for multicomponent alloy with grain size of 10–50μm following a multistep heat treatment cycle.     

基于模糊神经网络的地图匹配算法

提出了基于模糊神经网络的新的地图匹配算法.该算法综合了数字道路信息和GPS/DR定位信息,提取两个重要参数作为输入变量,即定位点到候选路段的投影距离及定位航向与候选路段方位角差.设计出了四层模糊神经网络及改进的收敛学习规则.实验结果表明所提出的算法能很好地匹配车辆行驶路段位置.      

基于BP神经网络的边坡稳定性分析

阐述了BP神经网络的基本思想、学习算法的步骤,以构建的学习样本为基础,建立边坡稳定性分析的BP神经网络模型,对学习样本进行归一化和训练,建立输入向量与输出向量的非线性关系,把训练好的网络运用于某露天矿边坡,结果表明：BP神经网训练结果与现场实际情况相符,说明该方法对工程实际有指导意义.     

Contract learning, clinical learning and clinicians

Current trends in education and training emphasise that learners, whether they are school children, students or adults, need to acquire generic skills and personal characteristics which will enable them to become independent self-directed learners. This will enable them to continue the process of learning throughout their lives. Recent recommendations for the reform of undergraduate medical education, for training of hospital doctors and general practitioners, and the higher profile now being given to continuing medical education, reflect the strength of this particular educational current sweeping through all levels of medical education. Learning contracts, developed through negotiation between a teacher and a learner, are especially effective educational tools for stimulating independent learning. This paper examines the theoretical basis of contract-learning and its relevance to clinical settings.     

Probability density methods for smooth function approximation and learning in populations of tuned spiking neurons

This article proposes a new method for interpreting computations performed by populations of spiking neurons. Neural firing is modeled as a rate-modulated random process for which the behavior of a neuron in response to external input can be completely described by its tuning function. I show that under certain conditions, cells with any desired tuning functions can be approximated using only spike coincidence detectors and linear operations on the spike output of existing cells. I show examples of adaptive algorithms based on only spike data that cause the underlying cell-tuning curves to converge according to standard supervised and unsupervised learning algorithms. Unsupervised learning based on principal components analysis leads to independent cell spike trains. These results suggest a duality relationship between the random discrete behavior of spiking cells and the deterministic smooth behavior of their tuning functions. Classical neural network approximation methods and learning algorithms based on continuous variables can thus be implemented within networks of spiking neurons without the need to make numerical estimates of the intermediate cell firing rates.     

Thickness and Shape Synthetical Adjustment for DC Mill Based on Dynamic Nerve-Fuzzy Control

Due to the complexity of thickness and shape synthetical adjustment system and the difficulties to build a mathematical model, a thickness and shape synthetical adjustment scheme on DC mill based on dynamic nerve-fuzzy control was put forward, and a self-organizing fuzzy control model was established. The structure of the network can be optimized dynamically. In the course of studying, the network can automatically adjust its structure based on the specific questions and make its structure the optimal. The input and output of the network are fuzzy sets, and the trained network can complete the composite relation, the fuzzy inference. For decreasing the off-line training time of BP network, the fuzzy sets are encoded. The simulation results indicate that the self-organizing fuzzy control based on dynamic neural network is better than traditional decoupling PID control.     

基于深度学习的热轧带钢头部厚度的命中预测

 针对热轧带钢头部厚度精度较低的问题,提出了一种基于深度学习的热轧带钢头部厚度的命中预测方法。在精轧过程中,带钢头部张力较小,且通常温度较低;同时轧机工艺参数复杂,精准设定存在困难,轧制带钢头部经常会出现厚度不合格的现象。利用深度神经网络的非线性拟合能力,设计带钢头部厚度预测模型,给轧机的参数设定提供参考、提高头部厚度命中率、减少钢材浪费。深度神经网络(DNN)包含输入层、隐藏层、输出层,使用TensorFlow开源机器学习框架设计预测模型并用程序实现。调整神经网络各参数,通过研究它们对模型性能的影响,优化预测模型。最后使用多种厚度的带钢测试数据训练并检验头部厚度预测模型,结果显示,分类预测命中准确率在80%以上。     

基于群体智能优化的MKL-SVM算法及肺结节识别

针对单核学习支持向量机无法兼顾学习能力与泛化能力以及多核函数参数寻优问题,提出了一种基于群体智能优化的多核学习支持向量机算法。首先,研究了五种单核函数对支持向量机分类性能的影响,进一步提出具有全局性质的多项式核和局部性质的拉普拉斯核凸组合形式的多核学习支持向量机算法;其次,为增加粒子多样性及快速寻优,将粒子群优化算法引入了遗传算法中的杂交操作,并用此改进的群体智能优化算法对多核学习支持向量机进行参数寻优。最后,分别采用深度特征与手工特征作为识别算法的输入,研究表明采用深度特征优于手工特征。故本文采用深度特征作为多核学习支持向量机的输入,以交叉遗传与粒子群混合智能优化算法作为其寻优方式。实验选取合作医院数据集对所提算法进行训练并初步测试,进一步为了验证所提算法的泛化能力,选取公开数据集LUNA16进行测试。实验结果表明,本文算法易于跳出局部最优解,提升了算法的学习能力与泛化能力,具有较优的分类性能。      

Cloud Neural Fuzzy PID Hybrid Integrated Algorithm of Flatness Control

In connection with the characteristics of multi-disturbance and nonlinearity of a system for flatness control in cold rolling process, a new intelligent PID control algorithm was proposed based on a cloud model, neural network and fuzzy integration. By indeterminacy artificial intelligence, the problem of fixing the membership functions of input variables and fuzzy rules was solved in an actual fuzzy system and the nonlinear mapping between variables was implemented by neural network. The algorithm has the adaptive learning ability of neural network and the indetermi- nacy of a cloud model in processing knowledge, which makes the fuzzy system have more persuasion in the process of knowledge inference, realizing the online adaptive regulation of PID parameters and avoiding the defects of the traditional PID controller. Simulation results show that the algorithm is simple, fast and robust with good control performance and application value.     

面向匹配决策问题的漏整合神经元稀疏ESN网络

为了对匹配决策问题进行建模与预测,提出了一种具有更多神经生理学特征的稀疏回声状态网络(ESN),并基于在线监督学习方法对网络进行训练.为了评估网络的匹配决策性能,设计了三组测试数据集对网络性能进行测试,并提出了一种基于网络期望输出与实际输出序列最大相关系数的评价方法.仿真结果表明,新模型只需要较少的训练时间即可获得较好的决策性能,且对发放时间间隔、平移和网络噪声具有较好的鲁棒性.      

Lexical Information Drives Perceptual Learning of Distorted Speech: Evidence From the Comprehension of Noise-Vocoded Sentences.

Speech comprehension is resistant to acoustic distortion in the input, reflecting listeners' ability to adjust perceptual processes to match the speech input. For noise-vocoded sentences, a manipulation that removes spectral detail from speech, listeners' reporting improved from near 0% to 70% correct over 30 sentences (Experiment 1). Learning was enhanced if listeners heard distorted sentences while they knew the identity of the undistorted target (Experiments 2 and 3). Learning was absent when listeners were trained with nonword sentences (Experiments 4 and 5), although the meaning of the training sentences did not affect learning (Experiment 5). Perceptual learning of noise-vocoded speech depends on higher level information, consistent with top-down, lexically driven learning. Similar processes may facilitate comprehension of speech in an unfamiliar accent or following cochlear implantation. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Neural Modeling of Square Surface Aerators

Applications of artificial neural networks in the field of aeration phenomena in surface aerators, which are not geometrically similar, are explored to predict reaeration rates under varying dynamic as well as geometric conditions. The primary network for prediction is a feed forward network with nonlinear elements. The network consists of an input layer, an output layer, a hidden layer, and the nonlinear transfer function in each processing element. The network requires supervised learning and the learning algorithm is the back-propagation. As back-propagation learning is affected by local minima, and to get over this aspect various other modifications have been suggested like Levenberg-Marquardt, quasi-Newton, conjugate-gradient, etc. The present study suggests that the Levenberg-Marquardt modification is a very efficient algorithm in comparison with others like quasi-Newton and conjugate-gradient. In the situations when the dimension of the input vector is large, and highly correlated, it is useful to reduce the dimension of the input vectors. An effective procedure for performing this operation is principal component analysis. The best prediction performance is achieved when the data are preprocessed using principal components analysis before they are fed to a back-propagated neural network, but at the cost of losing the physical significance of experimental data. The model thus developed can be used to predict the reaeration rate for different sizes of geometric elements (like rotor diameter, sizes of rotor, aerators’ geometry, water depth, etc.) under various dynamic conditions, i.e., the speed of the rotor.     

Unsupervised concept learning and value systematicitiy: A complex whole aids learning the parts.

Ease of learning new concepts may best be understood by simultaneously considering models of learning and theories of how "good" systems of categories are organized. The authors tested the effects on learning of value systematicity, a proposed organizing principle: If 1 attribute is predictive of another, it should predict still more. This principle derives from focused sampling in the internal feedback model (D. Billman & E. Heit, 1988) of unsupervised, or observational, learning. In 3 experiments, the authors tested how the organization of structure in input (value systematicity) affected unsupervised learning of categories about alien animals. Across all experiments, learning a target rule was easier in conditions with high value systematicity, relative to several low systematicity controls. The authors compare results to predictions of several learning models and consider the links between learning and the resulting category structure. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Avoidance learning of perceptual defense and vigilance.

Thirty-two undergraduates were randomly assigned to defense and vigilance training groups. "This study supports the view that perceptual defense and vigilance are learned reactions to anxiety arousing stimuli." A behavior theory analysis of the learning process is proposed. "According to this analysis, perceptual defense is learned when the perceptual response to a threatening stimulus is punished and competing perceptual responses are instrumental to anxiety reduction. Competing perceptual responses when reinforced are strengthened at the expense of the critical perceptual response. Perceptual vigilance is learned when the perceptual response to a threatening stimulus is reinforced by anxiety reduction and competing perceptual responses are punished." Learning for both groups "proceeded in the absence of awareness." (PsycINFO Database Record (c) 2010 APA, all rights reserved)