A quantitative solution to Constraint Satisfaction Problem (CSP)

Constraint Satisfaction Problem (CSP) is an important problem in artificial intelligence and operations research. Many practical problems can be formulated as CSP, i.e., finding a consistent value assignment to variables subject to a set of constraints. In this paper, we give a quantitative approach to solve the CSPs which deals uniformly with binary constraints as well as high order,k-ary (k ≥ 2) constraints. In this quantitative approach, using variable transformation and constraint transformation, a CSP is transformed into a satisfiability (SAT) problem. The SAT problem is then solved within a continuous search space. We will evaluate the performance of this method based on randomly generated SAT problem instances and regularly generatedk-ary (k ≥ 2) CSP problem instances.     

An artificial brain ATR's CAM-Brain Project aims to build/evolve an artificial brain with a million neural net modules inside a trillion cell Cellular Automata Machine

ATR's Evolutionary Systems Department aims to build (i.e. grow/evolve) an artificial brain by the year 2001. This artificial brain should initially contain thousands of interconnected artificial neural network modules, and be capable of controlling approximately 1000 “behaviors” in a “robot kitten”. The name given to this research project is “CAM-Brain”, because the neural networks (based on cellular automata) will be grown inside special hardware called Cellular Automata Machines (CAMs). Using a family of CAMs, each with its own processor to measure the performance quality or fitness of the evolved neural circuits, will allow the neural modules and their interconnections to be grown/evolved at electronic speeds. State of the art in CAM design is about 10 to the power 9 or 10 cells. Since a neural module of about 15 connected neurons can fit inside a cube of 100 cells on a side (1 million cells), a CAM which is specially adapted for CAM-Brain could contain thousands of interconnected modules, i.e. an artificial brain.     

Undecidable goals for completed acyclic programs

We prove for a well-known acyclic logic programP that it is undecidable whether or not a given goal is a logical consequence of the completion ofP. This complements recent decidability results for acyclic programs and bounded goals.     

Simple learning algorithms using divide and conquer

This paper investigates what happens when a learning algorithm for a classC attempts to learn target formulas from a different class. In many cases, the learning algorithm will find a bad attribute or a property of the target formula which precludes its membership in the classC. To continue the learning process, we proceed by building a decision tree according to the possible values of this attribute (divide) and recursively run the learning algorithm for each value (conquer). This paper shows how to recursively run the learning algorithm for each value using the oracles of the target.We demonstrate that the application of this idea on some known learning algorithms can both simplify the algorithm and provide additional power to learn more classes. In particular, we give a simple exact learning algorithm, using membership and equivalence queries, for the class of DNF that is almost unate, that is, unate with the addition ofO (logn) nonunate variables and a constant number of terms. We also find algorithms in different models for boolean functions that depend onk terms.     

Boosting imbalanced data learning with Wiener process oversampling

Learning from imbalanced data is a challenging task in a wide range of applications, which attracts significant research efforts from machine learning and data mining community. As a natural approach to this issue, oversampling balances the training samples through replicating existing samples or synthesizing new samples. In general, synthesization outperforms replication by supplying additional information on the minority class. However, the additional information needs to follow the same normal distribution of the training set, which further constrains the new samples within the predefined range of training set. In this paper, we present the Wiener process oversampling (WPO) technique that brings the physics phenomena into sample synthesization. WPO constructs a robust decision region by expanding the attribute ranges in training set while keeping the same normal distribution. The satisfactory performance of WPO can be achieved with much lower computing complexity. In addition, by integrating WPO with ensemble learning, the WPOBoost algorithm outperformsmany prevalent imbalance learning solutions.     

Iterative weighted least squares algorithms for neural networks classifiers

This paper discusses learning algorithms of layered neural networks from the standpoint of maximum likelihood estimation. At first we discuss learning algorithms for the most simple network with only one neuron. It is shown that Fisher information of the network, namely minus expected values of Hessian matrix, is given by a weighted covariance matrix of input vectors. A learning algorithm is presented on the basis of Fisher's scoring method which makes use of Fisher information instead of Hessian matrix in Newton's method. The algorithm can be interpreted as iterations of weighted least squares method. Then these results are extended to the layered network with one hidden layer. Fisher information for the layered network is given by a weighted covariance matrix of inputs of the network and outputs of hidden units. Since Newton's method for maximization problems has the difficulty when minus Hessian matrix is not positive definite, we propose a learning algorithm which makes use of Fisher information matrix, which is non-negative, instead of Hessian matrix. Moreover, to reduce the computation of full Fisher information matrix, we propose another algorithm which uses only block diagonal elements of Fisher information. The algorithm is reduced to an iterative weighted least squares algorithm in which each unit estimates its own weights by a weighted least squares method. It is experimentally shown that the proposed algorithms converge with fewer iterations than error back-propagation (BP) algorithm.     

结构化低秩字典学习的人脸识别

目的 现实中采集到的人脸图像通常受到光照、遮挡等环境因素的影响,使得同一类的人脸图像具有不同程度的差异性,不同类的人脸图像又具有不同程度的相似性,这极大地影响了人脸识别的准确性。为了解决上述问题对人脸识别造成的影响,在低秩矩阵恢复理论的基础上提出了具有识别力的结构化低秩字典学习的人脸识别算法。方法 该算法基于训练样本的标签信息将低秩正则化以及结构化稀疏同时引入到学习的具有识别力的字典上。在字典学习过程中,首先利用样本的重建误差约束样本与字典之间的关系;其次将Fisher准则应用到稀疏编码过程中,使其编码系数具有识别能力;由于训练样本中的噪声信息会影响字典的识别力,所以在低秩矩阵恢复理论的基础上将低秩正则化应用到字典学习过程中;接着,在字典学习过程中加入了结构化稀疏使其不丢失结构信息以保证对样本进行最优分类;最后再利用误差重构法对测试样本进行分类识别。结果 本文算法在AR以及ORL人脸数据库上分别进行了实验仿真。在AR人脸数据库中,为了分析样本不同维数对实验结果造成的影响,选取了第一时期拍摄的每人6幅图像,包括1幅围巾遮挡,2幅墨镜遮挡以及3幅脸部表情变化以及光照变化（未被遮挡）的图像作为训练样本,同时选取相同组合的样本图像作为测试样本,无论哪种方法,图像的维度越高识别率越高。对比SRC （sparse representation based on classification）算法与DKSVD （discriminative K-means singular value decomposition）算法的识别率可知,DKSVD算法通过字典学习减缓了训练样本中的不确定因素对识别结果的影响;对比DLRD_SR （discriminative low-rank dictionary learning for sparse representation）算法与FDDL （Fisher discriminative dictionary learning）算法的识别率可知,当图像有遮挡等噪声信息存在时,字典低秩化可以提高至少5.8%的识别率;对比本文算法与DLRD_SR算法可知,在字典学习的过程中加入Fisher准则后识别率显著提高,同时理想稀疏值能保证对样本进行最优的分类。当样本图像的维度达到500维时人脸图像在有围巾、墨镜遮挡的情况下识别率可达到85.2%;其中墨镜和围巾的遮挡程度分别可以看成是人脸图像的20%和40%,为了验证本文算法在不同脸部表情变化、光照改变以及遮挡情况下的有效性,根据训练样本的具体图像组合情况进行实验。无论哪种样本图像组合,本文算法在有遮挡存在的样本识别中具有显著优势。在训练样本只包含脸部表情变化、光照变化以及墨镜遮挡图像的情况下,本文算法的识别率高于其他算法至少2.7%,在训练样本只包含脸部表情变化、光照变化以及围巾遮挡图像的情况下,本文算法的识别率高于其他算法至少3.6%,在训练样本包含脸部表情变化、光照变化、围巾遮挡以及墨镜遮挡图像的情况下,其识别率高于其他算法至少1.9%。在ORL人脸数据库中,人脸图像在无遮挡的情况下识别率达到95.2%,稍低于FDDL算法的识别率;在随机块遮挡程度达到20%时,相比较于SRC算法、DKSVD算法、FDDL算法以及DLRD_SR算法,本文算法的识别率最高;当随机块遮挡程度达到50%时,以上算法的识别率均不高,但本文算法的其识别率仍然最高。结论 本文算法在人脸图像受到遮挡等因素的影响时具有一定的鲁棒性,实验结果表明该算法在人脸识别方面具有可行性。     

A study on effectiveness of extreme learning machine

Extreme learning machine (ELM), proposed by Huang et al., has been shown a promising learning algorithm for single-hidden layer feedforward neural networks (SLFNs). Nevertheless, because of the random choice of input weights and biases, the ELM algorithm sometimes makes the hidden layer output matrix H of SLFN not full column rank, which lowers the effectiveness of ELM. This paper discusses the effectiveness of ELM and proposes an improved algorithm called EELM that makes a proper selection of the input weights and bias before calculating the output weights, which ensures the full column rank of H in theory. This improves to some extend the learning rate (testing accuracy, prediction accuracy, learning time) and the robustness property of the networks. The experimental results based on both the benchmark function approximation and real-world problems including classification and regression applications show the good performances of EELM.     

A hybrid deep learning neural approach for emotion recognition from facial expressions for socially assistive robots

We have recently seen significant advancements in the development of robotic machines that are designed to assist people with their daily lives. Socially assistive robots are now able to perform a number of tasks autonomously and without human supervision. However, if these robots are to be accepted by human users, there is a need to focus on the form of human–robot interaction that is seen as acceptable by such users. In this paper, we extend our previous work, originally presented in Ruiz-Garcia et al. (in: Engineering applications of neural networks: 17th international conference, EANN 2016, Aberdeen, UK, September 2–5, 2016, proceedings, pp 79–93, 2016. https://doi.org/10.1007/978-3-319-44188-7_6), to provide emotion recognition from human facial expressions for application on a real-time robot. We expand on previous work by presenting a new hybrid deep learning emotion recognition model and preliminary results using this model on real-time emotion recognition performed by our humanoid robot. The hybrid emotion recognition model combines a Deep Convolutional Neural Network (CNN) for self-learnt feature extraction and a Support Vector Machine (SVM) for emotion classification. Compared to more complex approaches that use more layers in the convolutional model, this hybrid deep learning model produces state-of-the-art classification rate of \(96.26\%\), when tested on the Karolinska Directed Emotional Faces dataset (Lundqvist et al. in The Karolinska Directed Emotional Faces—KDEF, 1998), and offers similar performance on unseen data when tested on the Extended Cohn–Kanade dataset (Lucey et al. in: Proceedings of the third international workshop on CVPR for human communicative behaviour analysis (CVPR4HB 2010), San Francisco, USA, pp 94–101, 2010). This architecture also takes advantage of batch normalisation (Ioffe and Szegedy in Batch normalization: accelerating deep network training by reducing internal covariate shift. http://arxiv.org/abs/1502.03167, 2015) for fast learning from a smaller number of training samples. A comparison between Gabor filters and CNN for feature extraction, and between SVM and multilayer perceptron for classification is also provided.

     

Modelling and learning user preferences over sets

Although there has been significant research on modelling and learning user preferences for various types of objects, there has been relatively little work on the problem of representing and learning preferences over sets of objects. We introduce a representation language, DD-PREF, that balances preferences for particular objects with preferences about the properties of the set. Specifically, we focus on the depth of objects (i.e. preferences for specific attribute values over others) and on the diversity of sets (i.e. preferences for broad vs. narrow distributions of attribute values). The DD-PREF framework is general and can incorporate additional object- and set-based preferences. We describe a greedy algorithm, DD-Select, for selecting satisfying sets from a collection of new objects, given a preference in this language. We show how preferences represented in DD-PREF can be learned from training data. Experimental results are given for three domains: a blocks world domain with several different task-based preferences, a real-world music playlist collection, and rover image data gathered in desert training exercises.     

A genetic algorithm based robust learning credit assignment cerebellar model articulation controller

In this paper, a novel approach of genetic algorithm based robust learning credit assignment cerebellar model articulation controller (GCA-CMAC) is proposed. The cerebellar model articulation controller (CMAC) is a neurological model, which has an attractive property of learning speed. However, the distributions of errors into the addressed hypercubes of CMAC are not proportional to their credibility and may cause unacceptable learning performance. The credit assignment CMAC (CA-CMAC) can solve this problem by using the creditability of hypercubes that the calculated errors are assigned proportional to the inverse of learning times. Afterward, the obtained learning times can be optimized by genetic algorithm (GA) to increase its accuracy. In this paper, the proposed algorithm is to combine credit assignment ideas and GA to provide accurate learning for CMAC. Moreover, we embed the robust learning approach into the GCA-CMAC and dynamically adjust the learning constant for training data with noise or outliers. From simulation results, it shows that the proposed algorithm outperforms other CMACs.     

双Q网络学习的迁移强化学习算法

深度强化学习在训练过程中会探索大量环境样本,造成算法收敛时间过长,而重用或传输来自先前任务(源任务)学习的知识,对算法在新任务(目标任务)的学习具有提高算法收敛速度的潜力.为了提高算法学习效率,提出一种双Q网络学习的迁移强化学习算法,其基于actor-critic框架迁移源任务最优值函数的知识,使目标任务中值函数网络对策略作出更准确的评价,引导策略快速向最优策略方向更新.将该算法用于Open AI Gym以及在三维空间机械臂到达目标物位置的实验中,相比于常规深度强化学习算法取得了更好的效果,实验证明提出的双Q网络学习的迁移强化学习算法具有较快的收敛速度,并且在训练过程中算法探索更加稳定.     

A spectral algorithm for learning mixture models

We show that a simple spectral algorithm for learning a mixture of k spherical Gaussians in works remarkably well—it succeeds in identifying the Gaussians assuming essentially the minimum possible separation between their centers that keeps them unique (solving an open problem of Arora and Kannan (Proceedings of the 33rd ACM STOC, 2001). The sample complexity and running time are polynomial in both n and k. The algorithm can be applied to the more general problem of learning a mixture of “weakly isotropic” distributions (e.g. a mixture of uniform distributions on cubes).     

A Modified CMAC Algorithm Based on Credit Assignment

A Credit-Assignment CMAC (CA-CMAC) algorithm is proposed to reduce learning interference in conventional CMAC. In the proposed CA-CMAC, the error of the training sample distributed to the addressed memory cell is proportional to the cell's credibility, which is the inverse of the cell's activated times. The learning process of CA-CMAC is analyzed and conventional CMAC is proved to be a special case of CA-CMAC. Furthermore, the convergence properties of CA-CMAC both in batch learning and in incremental learning are investigated; meanwhile, the convergence theorems in the two learning schemes are obtained, respectively. Finally, simulations are carried out to testify the theorems and compare the performance of CA-CMAC with that of CMAC. Simulation results prove that CA-CMAC converges faster than conventional CMAC.     

Learning parities in the mistake-bound model

We study the problem of learning parity functions that depend on at most k variables (k-parities) attribute-efficiently in the mistake-bound model. We design a simple, deterministic, polynomial-time algorithm for learning k-parities with mistake bound . This is the first polynomial-time algorithm to learn ω(1)-parities in the mistake-bound model with mistake bound o(n).Using the standard conversion techniques from the mistake-bound model to the PAC model, our algorithm can also be used for learning k-parities in the PAC model. In particular, this implies a slight improvement over the results of Klivans and Servedio (2004) [1] for learning k-parities in the PAC model.We also show that the time algorithm from Klivans and Servedio (2004) [1] that PAC-learns k-parities with sample complexity can be extended to the mistake-bound model.     

DNA-based algorithms for learning Boolean formulae

We apply a DNA-based massively parallel exhaustive search to solving the computational learning problems of DNF (disjunctive normal form) Boolean formulae. Learning DNF formulae from examples is one of the most important open problems in computational learning theory and the problem of learning 3-term DNF formulae is known as intractable if RP NP. We propose new methods to encode any k-term DNF formula to a DNA strand, evaluate the encoded DNF formula for a truth-value assignment by using hybridization and primer extension with DNA polymerase, and find a consistent DNF formula with the given examples. By employing these methods, we show that the class of k-term DNF formulae (for any constant k) and the class of general DNF formulae are efficiently learnable on DNA computer.Second, in order for the DNA-based learning algorithm to be robust for errors in the data, we implement the weighted majority algorithm on DNA computers, called DNA-based majority algorithm via amplification (DNAMA), which take a strategy of ``amplifying' the consistent (correct) DNA strands. We show a theoretical analysis for the mistake bound of the DNA-based majority algorithm via amplification, and imply that the amplification to ``double the volumes' of the correct DNA strands in the test tube works well.     

主动学习的多标签图像在线分类

目的在多标签有监督学习框架中,构建具有较强泛化性能的分类器需要大量已标注训练样本,而实际应用中已标注样本少且获取代价十分昂贵。针对多标签图像分类中已标注样本数量不足和分类器再学习效率低的问题,提出一种结合主动学习的多标签图像在线分类算法。方法基于min-max理论,采用查询最具代表性和最具信息量的样本挑选策略主动地选择待标注样本,且基于KKT(Karush-Kuhn-Tucker)条件在线地更新多标签图像分类器。结果在4个公开的数据集上,采用4种多标签分类评价指标对本文算法进行评估。实验结果表明,本文采用的样本挑选方法比随机挑选样本方法和基于间隔的采样方法均占据明显优势;当分类器达到相同或相近的分类准确度时,利用本文的样本挑选策略选择的待标注样本数目要明显少于采用随机挑选样本方法和基于间隔的采样方法所需查询的样本数。结论本文算法一方面可以减少获取已标注样本所需的人工标注代价;另一方面也避免了传统的分类器重新训练时利用所有数据所产生的学习效率低下的问题,达到了当新数据到来时可实时更新分类器的目的。