Error backpropagation is an extremely effective algorithm for assigning credit in artificial neural networks. However, weight updates under Backprop depend on lengthy recursive computations and require separate output and error messages — features not shared by biological neurons, that are perhaps unnecessary. In this paper, we revisit Backprop and the credit assignment problem. We first decompose Backprop into a collection of interacting learning algorithms; provide regret bounds on the performance of these sub-algorithms; and factorize Backprop's error signals. Using these results, we derive a new credit assignment algorithm for nonparametric regression, Kickback, that is significantly simpler than Backprop. Finally, we provide a sufficient condition for Kickback to follow error gradients, and show that Kickback matches Backprop's performance on real-world regression benchmarks.

This paper proposes incremental preference elicitation methods for multiobjective state space search. Our approach consists in integrating weight elicitation and search to determine, in a vector-valued state-space graph, a solution path that best fits the Decision Maker's preferences. We first assume that the objective weights are imprecisely known and propose a state space search procedure to determine the set of possibly optimal solutions. Then, we introduce incremental elicitation strategies during the search that use queries to progressively reduce the set of admissible weights until a nearly-optimal path can be identified. The validity of our algorithms is established and numerical tests are provided to test their efficiency both in terms of number of queries and solution times.

Due to the openness and decentralization of the Web, mechanisms to represent and reason about the reliability of RDF data become essential. This paper embarks on a formal analysis of RDF data enriched with trust information by focusing on the characterization of its model-theoretic semantics and on the study of relevant reasoning problems. The impact of trust values on the computational complexity of well-known concepts related to the entailment of RDF graphs is studied. In particular, islands of tractability are identified for classes of acyclic and nearly-acyclic graphs. Moreover, an implementation of the framework and an experimental evaluation on real data are discussed.

The ability to determine patient acuity (or severity of illness) has immediate practical use for clinicians. We evaluate the use of multivariate timeseries modeling with the multi-task Gaussian process (GP) models using noisy, incomplete, sparse, heterogeneous and unevenly-sampled clinical data, including both physiological signals and clinical notes. The learned multi-task GP (MTGP) hyperparameters are then used to assess and forecast patient acuity. Experiments were conducted with two real clinical data sets acquired from ICU patients: firstly, estimating cerebrovascular pressure reactivity, an important indicator of secondary damage for traumatic brain injury patients, by learning the interactions between intracranial pressure and mean arterial blood pressure signals, and secondly, mortality prediction using clinical progress notes. In both cases, MTGPs provided improved results: an MTGP model provided better results than single-task GP models for signal interpolation and forecasting (0.91 vs 0.69 RMSE), and the use of MTGP hyperparameters obtained improved results when used as additional classification features (0.812 vs 0.788 AUC).

We describe an optimization-based approach for automatically creating well-edited movies from a 3D animation. While previous work has mostly focused on the problem of placing cameras to produce nice-looking views of the action, the problem of cutting and pasting shots from all available cameras has never been addressed extensively. In this paper, we review the main causes of editing errors in literature and propose an editing model relying on a minimization of such errors. We make a plausible semi-Markov assumption, resulting in a dynamic programming solution which is computationally efficient. We also show that our method can generate movies with different editing rhythms and validate the results through a user study. Combined with state-of-the-art cinematography, our approach therefore promises to significantly extend the expressiveness and naturalness of virtual movie-making.

We study convergence properties of iterative voting procedures. Such procedures are defined by a voting rule and a (restricted) iterative process, where at each step one agent can modify his vote towards a better outcome for himself. It is already known that if the iteration dynamics (the manner in which voters are allowed to modify their votes) are unrestricted, then the voting process may not converge. For most common voting rules this may be observed even under the best response dynamics limitation. It is therefore important to investigate whether and which natural restrictions on the dynamics of iterative voting procedures can guarantee convergence. To this end, we provide two general conditions on the dynamics based on iterative myopic improvements, each of which is sufficient for convergence. We then identify several classes of voting rules (including Positional Scoring Rules, Maximin, Copeland and Bucklin), along with their corresponding iterative processes, for which at least one of these conditions hold.

In this paper, we study protocols that allow to discern conscious and unconscious decisions of human beings; i.e., protocols that measure awareness. Consciousness is a central research theme in Neuroscience and AI, which remains, to date, an obscure phenomenon of human brains. Our starting point is a recent experiment, called Post Decision Wagering (PDW) (Persaud, McLeod, and Cowey 2007), that attempts to align experimenters' and subjects' objectives by leveraging financial incentives. We note a similarity with mechanism design, a research area which aims at the design of protocols that reconcile often divergent objectives through incentive-compatibility. We look at the issue of measuring awareness from this perspective. We abstract the setting underlying the PDW experiment and identify three factors that could make it ineffective: rationality, risk attitude and bias of subjects. Using mechanism design tools, we study the barrier between possibility and impossibility of incentive compatibility with respect to the aforementioned characteristics of subjects. We complete this study by showing how to use our mechanisms to potentially get a better understanding of consciousness.

We present a novel semantics for extracting bounded-level modules from RDF ontologies and databases augmented with safe inference rules, a la Datalog. Dealing with a recursive rule language poses challenging issues for defining the module semantics, and also makes module extraction algorithmically unsolvable in some cases. Our results include a set of module extraction algorithms compliant with the novel semantics. Experimental results show that the resulting framework is effective in extracting expressive modules from RDF datasets with formal guarantees, whilst controlling their succinctness.

Solution concepts from cooperative game theory, such as the Shapley value or the Banzhaf index, have recently been advocated as interesting extensions of standard measures of node centrality in networks. While this direction of research is promising, the computation of game-theoretic centrality can be challenging. In an attempt to address the computational issues of game-theoretic network centrality, we present a generic framework for constructing game-theoretic network centralities. We prove that all extensions that can be expressed in this framework are computable in polynomial time. Using our framework, we present the first game-theoretic extensions of weighted and normalized degree centralities, impact factor centrality,distance-scaled and normalized betweenness centrality,and closeness and normalized closeness centralities.

We introduce Extended Property Paths (EPPs), a significant enhancement of SPARQL property paths. EPPs allow to capture in a succinct way a larger class of navigational queries than property paths. We present the syntax and formal semantics of EPPs and introduce two different evaluation strategies. The first is based on an algorithm implemented in a custom query processor. The second strategy leverages a translation algorithm of EPPs into SPARQL queries that can be executed on existing SPARQL processors. We compare the two evaluation strategies on real data to highlight their pros and cons.