We propose a deep convolutional neural network architecture codenamed Inception that achieves the new state of the art for classification and detection in the ImageNet Large-Scale Visual Recognition Challenge 2014 (ILSVRC14). The main hallmark of this architecture is the improved utilization of the computing resources inside the network. By a carefully crafted design, we increased the depth and width of the network while keeping the computational budget constant. To optimize quality, the architectural decisions were based on the Hebbian principle and the intuition of multi-scale processing. One particular incarnation used in our submission for ILSVRC14 is called GoogLeNet, a 22 layers deep network, the quality of which is assessed in the context of classification and detection.

/pdf/going-deeper-with-convolutions-1yobw2o2ds.pdf

Going deeper with convolutions

There is, I think, something ethereal about i —the square root of minus one. I remember first hearing about it at school. It seemed an odd beast at that time—an intruder hovering on the edge of reality.

Usually familiarity dulls this sense of the bizarre, but in the case of i it was the reverse: over the years the sense of its surreal nature intensified. It seemed that it was impossible to write mathematics that described the real world in …

I and i

A major assumption in many machine learning and data mining algorithms is that the training and future data must be in the same feature space and have the same distribution. However, in many real-world applications, this assumption may not hold. For example, we sometimes have a classification task in one domain of interest, but we only have sufficient training data in another domain of interest, where the latter data may be in a different feature space or follow a different data distribution. In such cases, knowledge transfer, if done successfully, would greatly improve the performance of learning by avoiding much expensive data-labeling efforts. In recent years, transfer learning has emerged as a new learning framework to address this problem. This survey focuses on categorizing and reviewing the current progress on transfer learning for classification, regression, and clustering problems. In this survey, we discuss the relationship between transfer learning and other related machine learning techniques such as domain adaptation, multitask learning and sample selection bias, as well as covariate shift. We also explore some potential future issues in transfer learning research.

/pdf/a-survey-on-transfer-learning-1hjmu3otql.pdf

A Survey on Transfer Learning

Machine Learning is the study of methods for programming computers to learn. Computers are applied to a wide range of tasks, and for most of these it is relatively easy for programmers to design and implement the necessary software. However, there are many tasks for which this is difficult or impossible. These can be divided into four general categories. First, there are problems for which there exist no human experts. For example, in modern automated manufacturing facilities, there is a need to predict machine failures before they occur by analyzing sensor readings. Because the machines are new, there are no human experts who can be interviewed by a programmer to provide the knowledge necessary to build a computer system. A machine learning system can study recorded data and subsequent machine failures and learn prediction rules. Second, there are problems where human experts exist, but where they are unable to explain their expertise. This is the case in many perceptual tasks, such as speech recognition, hand-writing recognition, and natural language understanding. Virtually all humans exhibit expert-level abilities on these tasks, but none of them can describe the detailed steps that they follow as they perform them. Fortunately, humans can provide machines with examples of the inputs and correct outputs for these tasks, so machine learning algorithms can learn to map the inputs to the outputs. Third, there are problems where phenomena are changing rapidly. In finance, for example, people would like to predict the future behavior of the stock market, of consumer purchases, or of exchange rates. These behaviors change frequently, so that even if a programmer could construct a good predictive computer program, it would need to be rewritten frequently. A learning program can relieve the programmer of this burden by constantly modifying and tuning a set of learned prediction rules. Fourth, there are applications that need to be customized for each computer user separately. Consider, for example, a program to filter unwanted electronic mail messages. Different users will need different filters. It is unreasonable to expect each user to program his or her own rules, and it is infeasible to provide every user with a software engineer to keep the rules up-to-date. A machine learning system can learn which mail messages the user rejects and maintain the filtering rules automatically. Machine learning addresses many of the same research questions as the fields of statistics, data mining, and psychology, but with differences of emphasis. Statistics focuses on understanding the phenomena that have generated the data, often with the goal of testing different hypotheses about those phenomena. Data mining seeks to find patterns in the data that are understandable by people. Psychological studies of human learning aspire to understand the mechanisms underlying the various learning behaviors exhibited by people (concept learning, skill acquisition, strategy change, etc.).

Machine learning

Probability Distributions.- Linear Models for Regression.- Linear Models for Classification.- Neural Networks.- Kernel Methods.- Sparse Kernel Machines.- Graphical Models.- Mixture Models and EM.- Approximate Inference.- Sampling Methods.- Continuous Latent Variables.- Sequential Data.- Combining Models.

Pattern Recognition and Machine Learning

Ensemble learning aims to improve generalization ability by using multiple base learners. It is well-known that to construct a good ensemble, the base learners should be accurate as well as diverse. In this paper, unlabeled data is exploited to facilitate ensemble learning by helping augment the diversity among the base learners. Specifically, a semi-supervised ensemble method named UDEED is proposed. Unlike existing semi-supervised ensemble methods where error-prone pseudo-labels are estimated for unlabeled data to enlarge the labeled data to improve accuracy, UDEED works by maximizing accuracies of base learners on labeled data while maximizing diversity among them on unlabeled data. Experiments show that UDEED can effectively utilize unlabeled data for ensemble learning and is highly competitive to well-established semi-supervised ensemble methods.

Exploiting Unlabeled Data to Enhance Ensemble Diversity

Recently, there has been a growing research interest in the analysis of dynamic regret, which measures the performance of an online learner against a sequence of local minimizers. By exploiting the strong convexity, previous studies have shown that the dynamic regret can be upper bounded by the path-length of the comparator sequence. In this paper, we illustrate that the dynamic regret can be further improved by allowing the learner to query the gradient of the function multiple times, and meanwhile the strong convexity can be weakened to other non-degenerate conditions. Specifically, we introduce the squared path-length, which could be much smaller than the path-length, as a new regularity of the comparator sequence. When multiple gradients are accessible to the learner, we first demonstrate that the dynamic regret of strongly convex functions can be upper bounded by the minimum of the path-length and the squared path-length. We then extend our theoretical guarantee to functions that are semi-strongly convex or self-concordant. To the best of our knowledge, this is the first time that semi-strong convexity and self-concordance are utilized to tighten the dynamic regret.

Improved Dynamic Regret for Non-degenerate Functions

In traditional supervised classification, objects belong to only one class out of two or more disjoint classes. However in many real-world applications, objects may belong to more than one class at the same time. For example an article on the Greek debt crisis could among others belong to the following classes of a financial newspaper’s taxonomy: Greece, Eurozone, Economy and Markets. Such data are called multi-label. Early research on learning from multi-label data focused on automated document categorization (McCallum 1999; Schapire and Singer 2000), motivated by the need for low-cost annotation of large document collections maintained by news agencies, academic publishers and intellectual property organizations. As our ability to collect and store large amounts of digital content increased in recent years due to technological advances, so did the need for automated annotation of such content, leading to an increased interest in multi-label learning from the communities of image, video, music and multimedia information retrieval (Irie et al. 2010; Su et al. 2011; Lo et al. 2011; Yang and Chen 2011). Furthermore, applications of multi-label learning extend beyond automated content annotation to tag recommendation (Song et al. 2011), query categorization (González-Caro and Baeza-Yates 2011), gene function prediction (Valentini 2011), medical diagnosis (Taylor et al. 2010), drug discovery (Kawai and Takahashi 2009) and direct marketing (Zhang et al. 2006). Multi-label data are clearly ubiquitous.

/pdf/introduction-to-the-special-issue-on-learning-from-multi-57v3jl0onl.pdf

Introduction to the special issue on learning from multi-label data

Traditional face recognition systems attempt to achieve a high recognition accuracy, which implicitly assumes that the losses of all misclassifications are the same. However, in many real-world tasks this assumption is not always reasonable. For example, it will be troublesome if a face-recognition-based door-locker misclassifies a family member as a stranger such that s/he were not allowed to enter the house; but it will be a much more serious disaster if a stranger were misclassified as a family member and allowed to enter the house. In this paper, we propose a framework which formulates the problem as a multi-class cost-sensitive learning task, and propose a theoretically sound method based on Bayes decision theory to solve this problem. Experimental results demonstrate the effectiveness and efficiency of the proposed method.

https://cs.nju.edu.cn/zhouzh/zhouzh.files/publication/cvpr08.pdf

Cost-sensitive face recognition

In this paper, a fast adaptive neural network classifier named FANNC is proposed. FANNC exploits the advantages of both adaptive resonance theory and field theory. It needs only one-pass learning, and achieves not only high predictive accuracy but also fast learning speed. Besides, FANNC has incremental learning ability. When new instances are fed, it does not need to retrain the whole training set. Instead, it could learn the knowledge encoded in those instances through slightly adjusting the network topology when necessary, that is, adaptively appending one or two hidden units and corresponding connections to the existing network. This characteristic makes FANNC fit for real-time online learning tasks. Moreover, since the network architecture is adaptively set up, the disadvantage of manually determining the number of hidden units of most feed-forward neural networks is overcome. Benchmark tests show that FANNC is a preferable neural network classifier, which is superior to several other neural algorithms on both predictive accuracy and learning speed.

https://cs.nju.edu.cn/zhouzh/zhouzh.files/publication/kais00.pdf

Zhi-Hua Zhou

Papers

Exploiting Unlabeled Data to Enhance Ensemble Diversity

Improved Dynamic Regret for Non-degenerate Functions

Introduction to the special issue on learning from multi-label data

Cost-sensitive face recognition

FANNC: a fast adaptive neural network classifier