He then solves the simpler problem and uses the information obtained in the solution of the simpler problem in the solution of the original complex problem.

Methods of Approach At the outset it might be well to distinguish sharply between two general approaches to the problem of machine learning.

The editors have provided him with special introductory sections as a ing guide to a thorough understand of the reports.

Our descriptive vocabulary may still In this article I propose to describe an area of artificial contain proper names as modifiers but the explanatory intelligence (Al) research that I and several colleagues vocabulary now involves the impersonal qualities of an have been enaged in for a number of years.

This computer program accepts expressions in natural language as on-line input. It searches each expression for syntactic and semantic patterns. When a pattern match is discovered an appropriate reply is typed out in natural language so that a continuing dialogue develops between person and program. The dialogue is restricted to the context of interpersonal relations such as occurs in a psychiatric interview. The program is an interpreter/supervisor written in SUBALGOL and runs on a 32K IBM 7090 connected via a direct-data device to a PDP-1 and a Philco console.

Questions are asked and answered every day. Question answering (QA) technology aims to deliver the same facility online. It goes further than the more familiar search based on keywords (as in Google, Yahoo, and other search engines), in attempting to recognize what a question expresses and to respond with an actual answer. First, questions do not often translate into a simple list of keywords. For example, the question (1) Which countries did the pope visit in the 1960s? A much more complex set of keywords is needed in order to get anywhere close to the intended result, and experience shows that people will not learn how to formulate and use such sets. Second, QA takes responsibility for providing answers, rather than a searchable list of links to potentially relevant documents (web pages), highlighted by snippets of text that show how the query matched the documents. While this is not much of a burden when the answer appears in a snippet and further document access is unnecessary, QA technology aims to move this from being an accidental property of search to its focus. In keyword search and in much work to date on QA technology, the information seeking process has been seen as a one-shot affair: the user asks a question, and the system provides a satisfactory response. However, early work on QA (Section 1.1) did not make this assumption, and newly targeted applications are hindered by it: while a user may try to formulate a question whose answer is the information Question Answering 631 they want, they will not know whether they have succeeded until something has been returned for examination. If what is returned is unsatisfactory or, while not the answer, is still of interest, a user needs to be able to ask further questions that are understood in the context of the previous ones. For these target applications, QA must be part of a collaborative search process (Section 3.3). In the rest of this section, we give some historical background on QA systems (Section 1.1), on dialogue systems in which QA has played a significant role (Section 1.2), and on a particular QA task that has been a major driver of the field over the past 8 years (Section 1.3). Section 2 describes the current state of the art in QA systems, organized around the de facto architecture of such systems. Section 3 discusses some current directions in which QA is moving, including the development of interactive QA.

We address the problem of image collection summarization by learning mixtures of submodular functions. We argue that submodularity is very natural to this problem, and we show that a number of previously used scoring functions are submodular — a property not explicitly mentioned in these publications. We provide classes of submodular functions capturing the necessary properties of summaries, namely coverage, likelihood, and diversity. To learn mixtures of these submodular functions as scoring functions, we formulate summarization as a supervised learning problem using large-margin structured prediction. Furthermore, we introduce a novel evaluation metric, which we call V-ROUGE, for automatic summary scoring. While a similar metric called ROUGE has been successfully applied to document summarization [14], no such metric was known for quantifying the quality of image collection summaries. We provide a new dataset consisting of 14 real-world image collections along with many human-generated ground truth summaries collected using mechanical turk. We also extensively compare our method with previously explored methods for this problem and show that our learning approach outperforms all competitors on this new dataset. This paper provides, to our knowledge, the first systematic approach for quantifying the problem of image collection summarization, along with a new dataset of image collections and human summaries.

Kristen Grauman is an associate professor in the Department of Computer Science at the University of Texas at Austin. Her research in computer vision and machine learning focuses on visual search and object recognition. Before joining the University of Texas at Austin in 2007, she received her Ph.D. in the Electrical Engineering and Computer Science department at the Massachusetts Institute of Technology, in the Computer Science and Artificial Intelligence Laboratory. She is an Alfred P. Sloan Research Fellow and Microsoft Research New Faculty Fellow, a recipient of NSF CAREER and ONR Young Investigator awards, the Regents' Outstanding Teaching Award from the University of Texas System in 2012, the PAMI Young Researcher Award in 2013, the 2013 Computers and Thought Award from the International Joint Conference on Artificial Intelligence, and a Presidential Early Career Award for Scientists and Engineers (PECASE) in 2013. She and her collaborators were recognized with the CVPR Best Student Paper Award in 2008 for their work on hashing algorithms for large-scale image retrieval, and the Marr Best Paper Prize at ICCV in 2011 for their work on modeling relative visual attributes.

We present a novel approach to learn binary classifiers when only positive and unlabeled instances are available (PU learning). This problem is routinely cast as a supervised task with label noise in the negative set. We use an ensemble of SVM models trained on bootstrap resamples of the training data for increased robustness against label noise. The approach can be considered in a bagging framework which provides an intuitive explanation for its mechanics in a semi-supervised setting. We compared our method to state-of-the-art approaches in simulations using multiple public benchmark data sets. The included benchmark comprises three settings with increasing label noise: (i) fully supervised, (ii) PU learning and (iii) PU learning with false positives. Our approach shows a marginal improvement over existing methods in the second setting and a significant improvement in the third. Frank De Smet is a member of the medical management department of the National Alliance of Christian Mutualities. Accepted at Neurocomputing: SI on Advances in Learning with Label Noise 20/10/2014 1. Introduction Training binary classifiers on positive and unlabeled data is referred to as PU learning [31]. The absence of known negative training instances warrants appropriate learning methods. Inaccurate label information can be more problematic than attribute noise [45]. Specialised PU learning approaches are recommended when (i) negative labels cannot be acquired, (ii) the training data contains a large amount of false negatives or (iii) the positive set has many outliers. Practical applications of PU learning typically feature large, imbalanced training sets with a small amount of labeled (positive) and a large amount of unlabeled training instances. The PU learning problem arises in various settings, including web page classification [44], intrusion detection [26] and bioinformatics tasks such as variant prioritization [42], gene prioritization [1, 35] and virtual screening of drug compounds [41]. Though these applications share a common underlying learning problem, the final evaluation criteria may be fundamentally different.

Participants Intelligence (AAAI-15) and the Twenty-Seventh Conference in the AAAI-15 Robotics Exhibition and the on Innovative Applications of Artificial Intelligence AAAI-15 Video Competition are encouraged to contribute (IAAI-15) will be held January 25-29 at the to the Demonstration Program with their systems, Hyatt Regency Austin in Austin, Texas, USA. AAAI is working October 8 (Papers Due) closely with the local AI community to create opportunities The Senior Member Track provides an opportunity for attendees to experience AI in Texas! Attendees for established researchers in the AI community to can also enjoy nearly 200 music venues that feature give a broad talk on a well-developed body of everything from rock and blues to country and research, an important new research area, or a promising jazz every night of the week. Austin cuisine has new topic. This year, new "Blue Sky Ideas" track expanded from barbecue and Tex-Mex to award-winning is seeking presentations aimed at presenting ideas and inventive international cuisine, and blossomed and visions that can stimulate the research community beyond brick-and-mortar restaurants to a to pursue new directions, such as new problems, vibrant, citywide food truck movement.