Search

In this paper, we study the formalism of unsupervised multi-class domain adaptation (multi-class UDA), which underlies a few recent algorithms whose learning objectives are only motivated empirically. Multi-Class Scoring Disagreement (MCSD) divergence is presented by aggregating the absolute margin violations in multi-class classification, and this proposed MCSD is able to fully characterize the relations between any pair of multi-class scoring hypotheses. By using MCSD as a measure of domain distance, we develop a new domain adaptation bound for multi-class UDA; its data-dependent, probably approximately correct bound is also developed that naturally suggests adversarial learning objectives to align conditional feature distributions across source and target domains. Consequently, an algorithmic framework of Multi-class Domain-adversarial learning Networks (McDalNets) is developed, and its different instantiations via surrogate learning objectives either coincide with or resemble a few recently popular methods, thus (partially) underscoring their practical effectiveness. Based on our identical theory for multi-class UDA, we also introduce a new algorithm of Domain-Symmetric Networks (SymmNets), which is featured by a novel adversarial strategy of domain confusion and discrimination. SymmNets affords simple extensions that work equally well under the problem settings of either closed set, partial, or open set UDA. We conduct careful empirical studies to compare different algorithms of McDalNets and our newly introduced SymmNets. Experiments verify our theoretical analysis and show the efficacy of our proposed SymmNets. In addition, we have made our implementation code publicly available. [ABSTRACT FROM AUTHOR]

Dictionary learning has achieved remarkable success on a wide range of machine learning-based applications. In this paper, a locality-adaptive structured dictionary learning (LASDL) algorithm for cross-domain recognition is proposed. In the LASDL, a projective structured double reconstruction strategy is developed to train class-oriented sub-dictionaries from the specific classes of cross-domain samples. The strategy benefits to make full advantage of the discriminative information of cross-domain data and bridge the distribution divergence between two different domains. Meanwhile, an adaptive geometrical structure preserving function is designed to not only preserve the local manifold structures spanned by the representation coefficient spaces of the source and target domains, but also impose a constraint that the coefficients should keep closer to their class centers, which is propitious to reduce the distribution divergence and make the representation more accurate. With the structured linear coding technique, the final cross-domain recognition can be efficiently performed by determining class-specific reconstruction error. The optimization of the proposed LASDL model can be efficiently solved by simple least square method and alternating direction method of multipliers (ADMM) algorithm. Extensive experimental results validate the superiority of the proposed algorithm in contrast to other state-of-the-art predecessors. [ABSTRACT FROM AUTHOR]

In this paper, we investigate designs of linear precoders for vector Gaussian channels via stochastic optimizations and deep neural networks (DNNs). We assume that the channel inputs are drawn from practical finite alphabets, and we search for precoders maximizing the mutual information between channel inputs and outputs. Though the problem is generally non-convex, we prove that when the right singular matrix of precoder is fixed, any local optima of this problem is a global optima. Based on this fact, an efficient projected stochastic gradient descent (PSGD) algorithm is designed to search the optimal precoders. Moreover, to reduce the complexity of calculating a posterior means involved in gradients calculation, K-best algorithm is adopted to make approximations of a posterior means with negligible loss of accuracy. Furthermore, to avoid explicit calculation of mutual information and its gradients, DNN-based autoencoders (AEs) are constructed for this precoding task, and an efficient training algorithm is proposed. We also prove that the AEs, with ‘softmax’ activation function and ‘categorical cross entropy’ loss, maximize the mutual information under reasonable assumptions. Then, in order to extend the AE methods to large scale systems, ‘sigmoid’ activation function and ‘binary cross entropy’ loss are used such that the size of AEs will not grow prohibitively large. We prove that this maximizes a lower bound of the mutual information under reasonable assumptions. Finally, to make the precoders practical for high speed wireless scenarios, we propose an offline training paradigm which trains DNNs to infer optimal precoders given channel state information instead of training online for every different channel. Simulation results show that all the proposed methods work well in maximizing mutual information and improving bit error rate (BER) performance. [ABSTRACT FROM AUTHOR]

In real-world recognition/classification tasks, limited by various objective factors, it is usually difficult to collect training samples to exhaust all classes when training a recognizer or classifier. A more realistic scenario is open set recognition (OSR), where incomplete knowledge of the world exists at training time, and unknown classes can be submitted to an algorithm during testing, requiring the classifiers to not only accurately classify the seen classes, but also effectively deal with unseen ones. This paper provides a comprehensive survey of existing open set recognition techniques covering various aspects ranging from related definitions, representations of models, datasets, evaluation criteria, and algorithm comparisons. Furthermore, we briefly analyze the relationships between OSR and its related tasks including zero-shot, one-shot (few-shot) recognition/learning techniques, classification with reject option, and so forth. Additionally, we also review the open world recognition which can be seen as a natural extension of OSR. Importantly, we highlight the limitations of existing approaches and point out some promising subsequent research directions in this field. [ABSTRACT FROM AUTHOR]

Multi-label classification is an important research topic in machine learning, for which exploiting label dependencies is an effective modeling principle. Recently, probabilistic models have shown great potential in discovering dependencies among labels. In this paper, motivated by the recent success of multi-view learning to improve the generalization performance, we propose a novel multi-view probabilistic model named latent conditional Bernoulli mixture (LCBM) for multi-label classification. LCBM is a generative model taking features from different views as inputs, and conditional on the latent subspace shared by the views a Bernoulli mixture model is adopted to build label dependencies. Inside each component of the mixture, the labels have a weak correlation which facilitates computational convenience. The mean field variational inference framework is used to carry out approximate posterior inference in the probabilistic model, where we propose a Gaussian mixture variational autoencoder (GMVAE) for effective posterior approximation. We further develop a scalable stochastic training algorithm for efficiently optimizing the model parameters and variational parameters, and derive an efficient prediction procedure based on greedy search. Experimental results on multiple benchmark datasets show that our approach outperforms other state-of-the-art methods under various metrics. [ABSTRACT FROM AUTHOR]

Deep neural networks (DNNs) have been widely applied for classification tasks in wireless communications such as link scheduling, user association and base station (BS) muting. When training a DNN using datasets with highly-skewed class distribution where most data belong to a few majority classes, learning performance for minority classes will degrade since the imbalanced data forces the training process to be biased towards the majority classes. In many wireless problems, multiple decisions for classification are made jointly, say deciding which BSs in a network should be muted to avoid interference. In this paper, we employ DNN to learn an optimal predictive interference coordination policy, and strive to avoid biased learning in such a multi-label multi-class classification problem. The major contribution is proposing a training method to encourage fairness among classes by minimizing the maximal cost of decisions among classes, which is converted into a problem to optimize the weighting factors on the training cost of each class. We provide a gradient descent algorithm to optimize the weighting factors and the model parameters of the DNN by alternative updates. Simulation results show that the training method can improve the learning performance for the minority class and can achieve higher network utility than existing training methods. [ABSTRACT FROM AUTHOR]

As the training dataset size and the model size of machine learning increase rapidly, more computing resources are consumed to speedup the training process. However, the scalability and performance reproducibility of parallel machine learning training, which mainly uses stochastic optimization algorithms, are limited. In this paper, we demonstrate that the sample difference in the dataset plays a prominent role in the scalability of parallel machine learning algorithms. We propose to use statistical properties of dataset to measure sample differences. These properties include the variance of sample features, sample sparsity, sample diversity, and similarity in sampling sequences. We choose four types of parallel training algorithms as our research objects: (1) the asynchronous parallel SGD algorithm (Hogwild! algorithm), (2) the parallel model average SGD algorithm (minibatch SGD algorithm), (3) the decentralization optimization algorithm, and (4) the dual coordinate optimization (DADM algorithm). Our results show that the statistical properties of training datasets determine the scalability upper bound of these parallel training algorithms. [ABSTRACT FROM AUTHOR]