This report summarizes the conference and related activities. For the first time in AIIDE's history, the main program of the conference was preceded by three workshops: Intelligent Narrative Technologies workshop, the workshop on Nonplayer Character AI, and the Artificial Intelligence in the Game Design Process workshop. All three attracted a substantial audience and led to exciting debates and fruitful discussions (figure 1). In total, 24 papers were presented in the three workshops. The Intelligent Narrative Technologies workshop included papers on story representation, dialogue generation, narrative visualization, and authoring interfaces for interactive narrative, and a panel on corpus-based approaches to modeling narrative.

On the day before the exhibition the participants convened a workshop of 18 short talks. Each track's exhibitors presented a summary of their exhibit. In addition, four guest speakers provided a broader context for all of the exhibitors' efforts. The first guest speaker was the National Science Foundation's Sven Koenig, who highlighted several federal programs that support projects in embodied intelligence. Koenig also provided insights into some of these program's specific priorities, such as international collaborations and educational engagement.

Of course, this is far from the first attempt to plot a course toward human-level AGI: arguably this was the goal of the founders of the field of artificial intelligence in the 1950s, and has been pursued by a steady stream of AI researchers since, even as the majority of the AI field has focused its attention on more narrow, specific subgoals. The ideas presented here build on the ideas of others in innumerable ways, but to review the history of AI and situate the current effort in the context of its predecessors would require a much longer article than this one. Thus we have chosen to focus on the results of our AGI roadmap discussions, acknowledging in a broad way the many debts owed to many prior researchers. References to the prior literature on evaluation of advanced AI systems are given by Laird (Laird et al. 2009) and Geortzel and Bugaj (2009), which may in a limited sense be considered prequels to this article. We begin by discussing AGI in general and adopt a pragmatic goal for measuring progress toward its attainment. An initial capability landscape for AGI The heterogeneity of general intelligence in will be presented, drawing on major themes from humans makes it practically impossible to develop developmental psychology and illuminated by a comprehensive, fine-grained measurement system mathematical, physiological, and informationprocessing for AGI. While we encourage research in defining perspectives. The challenge of identifying such high-fidelity metrics for specific capabilities, appropriate tasks and environments for measuring we feel that at this stage of AGI development AGI will be taken up. Several scenarios will a pragmatic, high-level goal is the best we can be presented as milestones outlining a roadmap agree upon. I advocate beginning with a system that has minimal, although extensive, built-in capabilities. Many variant approaches have been proposed A classic example of the narrow AI approach was for achieving such a goal, and both the AI and AGI IBM's Deep Blue system (Campbell, Hoane, and communities have been working for decades on Hsu 2002), which successfully defeated world chess the myriad subgoals that would have to be champion Gary Kasparov but could not readily achieved and integrated to deliver a comprehensive apply that skill to any other problem domain without AGI system.

Machine learning research is often conducted in vitro, divorced from motivating practical applications. A researcher might develop a new method for the general task of classification, then assess its utility by comparing its performance (such as accuracy or AUC) to that of existing classification models on publicly available datasets. In terms of advancing machine learning as an academic discipline, this approach has thus far proven quite fruitful. However, it is our view that the most interesting open problems in machine learning are those that arise during its application to real-world problems. We illustrate this point by reviewing two of our interdisciplinary collaborations, both of which have posed unique machine learning problems, providing fertile ground for novel research.

Collaborative filtering is a rapidly advancing research area. Every year several new techniques are proposed and yet it is not clear which of the techniques work best and under what conditions. In this paper we conduct a study comparing several collaborative filtering techniques -- both classic and recent state-of-the-art -- in a variety of experimental contexts. Specifically, we report conclusions controlling for number of items, number of users, sparsity level, performance criteria, and computational complexity. Our conclusions identify what algorithms work well and in what conditions, and contribute to both industrial deployment collaborative filtering algorithms and to the research community.

Orseau and Ring, as well as Dewey, have recently described problems, including self-delusion, with the behavior of agents using various definitions of utility functions. An agent's utility function is defined in terms of the agent's history of interactions with its environment. This paper argues, via two examples, that the behavior problems can be avoided by formulating the utility function in two steps: 1) inferring a model of the environment from interactions, and 2) computing utility as a function of the environment model. Basing a utility function on a model that the agent must learn implies that the utility function must initially be expressed in terms of specifications to be matched to structures in the learned model. These specifications constitute prior assumptions about the environment so this approach will not work with arbitrary environments. But the approach should work for agents designed by humans to act in the physical world. The paper also addresses the issue of self-modifying agents and shows that if provided with the possibility to modify their utility functions agents will not choose to do so, under some usual assumptions.

A major benefit of graphical models is that most knowledge is captured in the model structure. Many models, however, produce inference problems with a lot of symmetries not reflected in the graphical structure and hence not exploitable by efficient inference techniques such as belief propagation (BP). In this paper, we present a new and simple BP algorithm, called counting BP, that exploits such additional symmetries. Starting from a given factor graph, counting BP first constructs a compressed factor graph of clusternodes and clusterfactors, corresponding to sets of nodes and factors that are indistinguishable given the evidence. Then it runs a modified BP algorithm on the compressed graph that is equivalent to running BP on the original factor graph. Our experiments show that counting BP is applicable to a variety of important AI tasks such as (dynamic) relational models and boolean model counting, and that significant efficiency gains are obtainable, often by orders of magnitude.

The problem of joint feature selection across a group of related tasks has applications in many areas including biomedical informatics and computer vision. We consider the l2,1-norm regularized regression model for joint feature selection from multiple tasks, which can be derived in the probabilistic framework by assuming a suitable prior from the exponential family. One appealing feature of the l2,1-norm regularization is that it encourages multiple predictors to share similar sparsity patterns. However, the resulting optimization problem is challenging to solve due to the non-smoothness of the l2,1-norm regularization. In this paper, we propose to accelerate the computation by reformulating it as two equivalent smooth convex optimization problems which are then solved via the Nesterov's method-an optimal first-order black-box method for smooth convex optimization. A key building block in solving the reformulations is the Euclidean projection. We show that the Euclidean projection for the first reformulation can be analytically computed, while the Euclidean projection for the second one can be computed in linear time. Empirical evaluations on several data sets verify the efficiency of the proposed algorithms.

We demonstrate that a number of sociology models for social network dynamics can be viewed as continuous time Bayesian networks (CTBNs). A sampling-based approximate inference method for CTBNs can be used as the basis of an expectation-maximization procedure that achieves better accuracy in estimating the parameters of the model than the standard method of moments algorithmfromthe sociology literature. We extend the existing social network models to allow for indirect and asynchronous observations of the links. A Markov chain Monte Carlo sampling algorithm for this new model permits estimation and inference. We provide results on both a synthetic network (for verification) and real social network data.

We present an objective function for learning with unlabeled data that utilizes auxiliary expectation constraints. We optimize this objective function using a procedure that alternates between information and moment projections. Our method provides an alternate interpretation of the posterior regularization framework (Graca et al., 2008), maintains uncertainty during optimization unlike constraint-driven learning (Chang et al., 2007), and is more efficient than generalized expectation criteria (Mann & McCallum, 2008). Applications of this framework include minimally supervised learning, semisupervised learning, and learning with constraints that are more expressive than the underlying model. In experiments, we demonstrate comparable accuracy to generalized expectation criteria for minimally supervised learning, and use expressive structural constraints to guide semi-supervised learning, providing a 3%-6% improvement over stateof-the-art constraint-driven learning.