With the advances in distributed computation, machine learn-ing and deep neural networks, we enter into an era that it is possible to build a real world image recognition system. There are three essential components to build a real-world image recognition system: 1) creating representative features, 2) de-signing powerful learning approaches, and 3) identifying massive training data. While extensive researches have been done on the first two aspects, much less attention has been paid on the third. In this paper, we present an end-to-end Web knowledge discovery system, Prajna. Starting from an arbi-trary set of entities as inputs, Prajna automatically crawls im-ages from multiple sources, identifies images that have relia-bly labeled, trains models and build a recognition system that is capable of recognizing any new images of the entity set. Due to the high cost of manual data labeling, leveraging the massive yet noisy data on the Internet is a natural idea, but the practical engineering aspect is highly challenging. Prajna fo-cuses on separating reliable training data from extensive noisy data, which is a key to the capability of extending an image recognition system to support arbitrary entities. In this paper, we will analyze the intrinsic characteristics of Internet image data, and find ways to mine accurate and informative infor-mation from those data to build a training set, which is then used to train image recognition models. Prajna is capable of automatically building an image recognition system for those entities as long as we can collect sufficient number of images of the entities on the Web.

Large-scale distributed computing has made available the resources necessary to solve "AI-hard" problems. As a result, it becomes feasible to automate the processing of such problems, but accuracy is not very high due to the conceptual difficulty of these problems. In this paper, we integrated crowdsourcing with MapReduce to provide a scalable innovative human-machine solution to AI-hard problems, which is called CrowdMR. In CrowdMR, the majority of problem instances are automatically processed by machine while the troublesome instances are redirected to human via crowdsourcing. The results returned from crowdsourcing are validated in the form of CAPTCHA (Completely Automated Public Turing test to Tell Computers and Humans Apart) before adding to the output. An incremental scheduling method was brought forward to combine the results from machine and human in a "pay-as-you-go" way.

This paper is concerned with knowledge compilation (KC), a family of approaches developed in AI for more than twenty years. Knowledge compilation consists in pre-processing some pieces of the available information in order to improve the computational efficiency (especially, the time complexity) of some tasks. In this paper, the focus is laid on three KC topics which gave rise to many works: the development of knowledge compilation techniques for the clausal entailment problem in propositional logic, the concept of compilability and the notion of knowledge compilation map. The three topics, as well as an overview of the main results from the literature, are presented. Some recent research lines are also discussed.

Applications of graphical models often require the use of approximate inference, such as sequential importance sampling (SIS), for estimation of the model distribution given partial evidence, i.e., the target distribution. However, when SIS proposal and target distributions are dissimilar, such procedures lead to biased estimates or require a prohibitive number of samples. We introduce ReBaSIS, a method that better approximates the target distribution by sampling variable by variable from existing importance samplers and accepting or rejecting each proposed assignment in the sequence: a choice made based on anticipating upcoming evidence. We relate the per-variable proposal and model distributions by expected weight ratios of sequence completions and show that we can learn accurate models of optimal acceptance probabilities from local samples. In a continuous-time domain, our method improves upon previous importance samplers by transforming an SIS problem into a machine learning one.

Motivated by problems such as molecular energy prediction, we derive an (improper) kernel between geometric inputs, that is able to capture the relevant rotational and translation invariances in geometric data. Since many physical simulations based upon geometric data produce derivatives of the output quantity with respect to the input positions, we derive an approach that incorporates derivative information into our kernel learning. We further show how to exploit the low rank structure of the resulting kernel matrices to speed up learning. Finally, we evaluated the method in the context of molecular energy prediction, showing good performance for modeling previously unseen molecular configurations. Integrating the approach into a Bayesian optimization, we show substantial improvement over the state of the art in molecular energy optimization.

Matching and merging data from conflicting sources is the bread and butter of data integration, which drives search verticals, e-commerce comparison sites and cyber intelligence. Schema matching lifts data integration - traditionally focused on well-structured data - to highly heterogeneous sources. While schema matching has enjoyed significant success in matching data attributes, inconsistencies can exist at a deeper level, making full integration difficult or impossible. We propose a more fine-grained approach that focuses on correspondences between the values of attributes across data sources. Since the semantics of attribute values derive from their use and co-occurrence, we argue for the suitability of canonical correlation analysis (CCA) and its variants. We demonstrate the superior statistical and computational performance of multiple sparse CCA compared to a suite of baseline algorithms, on two datasets which we are releasing to stimulate further research. Our crowd-annotated data covers both cases that are relatively easy for humans to supply ground-truth, and that are inherently difficult for human computation.

Combinatory Categorial Grammar (CCG) is a lexicalized grammar formalism in which words are associated with categories that, in combination with a small universal set of rules, specify the syntactic configurations in which they may occur. Categories are selected from a large, recursively-defined set; this leads to high word-to-category ambiguity, which is one of the primary factors that make learning CCG parsers difficult, especially in the face of little data. Previous work has shown that learning sequence models for CCG tagging can be improved by using linguistically-motivated prior probability distributions over potential categories. We extend this approach to the task of learning a CCG parser from weak supervision. We present a Bayesian formulation for CCG parser induction that assumes only supervision in the form of an incomplete tag dictionary mapping some word types to sets of potential categories. Our approach outperforms a baseline model trained with uniform priors by exploiting universal, intrinsic properties of the CCG formalism to bias the model toward simpler, more cross-linguistically common categories.

Salient object detection has gradually become a popular topic in robotics and computer vision research. This paper presents a real-time system that detects salient object by integrating objectness, foreground and compactness measures. Our algorithm consists of four basic steps. First, our method generates the objectness map via object proposals. Based on the objectness map, we estimate the background margin and compute the corresponding foreground map which prefers the foreground objects. From the objectness map and the foreground map, the compactness map is formed to favor the compact objects. We then integrate those cues to form a pixel-accurate saliency map which covers the salient objects and consistently separates fore- and background.

Multi-label image classification is of significant interest due to its major role in real-world web image analysis applications such as large-scale image retrieval and browsing. Recently, matrix completion (MC) has been developed to deal with multi-label classification tasks. MC has distinct advantages, such as robustness to missing entries in the feature and label spaces and a natural ability to handle multi-label problems. However, current MC-based multi-label image classification methods only consider data represented by a single-view feature, therefore, do not precisely characterize images that contain several semantic concepts. An intuitive way to utilize multiple features taken from different views is to concatenate the different features into a long vector; however, this concatenation is prone to over-fitting and leads to high time complexity in MC-based image classification. Therefore, we present a novel multi-view learning model for MC-based image classification, called low-rank multi-view matrix completion (lrMMC), which first seeks a low-dimensional common representation of all views by utilizing the proposed low-rank multi-view learning (lrMVL) algorithm. In lrMVL, the common subspace is constrained to be low rank so that it is suitable for MC. In addition, combination weights are learned to explore complementarity between different views. An efficient solver based on fixed-point continuation (FPC) is developed for optimization, and the learned low-rank representation is then incorporated into MC-based image classification. Extensive experimentation on the challenging PASCAL VOC' 07 dataset demonstrates the superiority of lrMMC compared to other multi-label image classification approaches.

Online learning algorithms require to often recompute least squares regression estimates of parameters. We study improving the computational complexity of such algorithms by using stochastic gradient descent (SGD) type schemes in place of classic regression solvers. We show that SGD schemes efficiently track the true solutions of the regression problems, even in the presence of a drift. This finding coupled with an $O(d)$ improvement in complexity, where $d$ is the dimension of the data, make them attractive for implementation in the \textit{big data} settings. In the case when strong convexity in the regression problem is guaranteed, we provide bounds on the error both in expectation and high probability (the latter is often needed to provide theoretical guarantees for higher level algorithms), despite the drifting least squares solution. As an example of this case we prove that the regret performance of an SGD version of the PEGE linear bandit algorithm is worse than that of PEGE itself only by a factor of $O(\log^4 n)$. When strong convexity of the regression problem cannot be guaranteed, we investigate using an adaptive regularisation. We make an empirical study of an adaptively regularised, SGD version of LinUCB in a news article recommendation application, which uses the large scale news recommendation dataset from Yahoo! front page. These experiments show a large gain in computational complexity and a consistently low tracking error.