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.

Knowledge bases have the potential to advance artificial intelligence, but often suffer from recall problems, i.e., lack of knowledge of new entities and relations. On the contrary, social media such as Twitter provide abundance of data, in a timely manner: information spreads at an incredible pace and is posted long before it makes it into more commonly used resources for knowledge extraction. In this paper we address the question whether we can exploit social media to extract new facts, which may at first seem like finding needles in haystacks. We collect tweets about 60 entities in Freebase and compare four methods to extract binary relation candidates, based on syntactic and semantic parsing and simple mechanism for factuality scoring. The extracted facts are manually evaluated in terms of their correctness and relevance for search. We show that moving from bottom-up syntactic or semantic dependency parsing formalisms to top-down frame-semantic processing improves the robustness of knowledge extraction, producing more intelligible fact candidates of better quality. In order to evaluate the quality of frame semantic parsing on Twitter intrinsically, we make a multiply frame-annotated dataset of tweets publicly available.

Querying inconsistent ontologies is an intriguing new problem that gave rise to a flourishing research activity in the description logic (DL) community. The computational complexity of consistent query answering under the main DLs is rather well understood; however, little is known about existential rules. The goal of the current work is to perform an in-depth analysis of the complexity of consistent query answering under the main decidable classes of existential rules enriched with negative constraints. Our investigation focuses on one of the most prominent inconsistency-tolerant semantics, namely, the AR semantics. We establish a generic complexity result, which demonstrates the tight connection between classical and consistent query answering. This result allows us to obtain in a uniform way a relatively complete picture of the complexity of our problem.

Constraint programming is a natural paradigm for many combinatorial optimisation problems. The complexity of constraint satisfaction for various forms of constraints has been widely-studied, both to inform the choice of appropriate algorithms, and to understand better the boundary between polynomial-time complexity and NP-hardness. In constraint programming it is well-known that any constraint satisfaction problem can be converted to an equivalent binary problem using the so-called dual encoding. Using this standard approach any fixed collection of constraints, of arbitrary arity, can be converted to an equivalent set of constraints of arity at most two. Here we show that this transformation, although it changes the domain of the constraints, preserves all the relevant algebraic properties that determine the complexity. Moreover, we show that the dual encoding preserves many of the key algorithmic properties of the original instance. We also show that this remains true for more general valued constraint languages, where constraints may assign different cost values to different assignments. Hence, we obtain a simple proof of the fact that to classify the computational complexity of all valued constraint languages it suffices to classify only binary valued constraint languages.

We recently have witnessed many ground-breaking results in machine learning and computer vision, generated by using deep convolutional neural networks (CNN). While the success mainly stems from the large volume of training data and the deep network architectures, the vector processing hardware (e.g. GPU) undisputedly plays a vital role in modern CNN implementations to support massive computation. Though much attention was paid in the extent literature to understand the algorithmic side of deep CNN, little research was dedicated to the vectorization for scaling up CNNs. In this paper, we studied the vectorization process of key building blocks in deep CNNs, in order to better understand and facilitate parallel implementation. Key steps in training and testing deep CNNs are abstracted as matrix and vector operators, upon which parallelism can be easily achieved. We developed and compared six implementations with various degrees of vectorization with which we illustrated the impact of vectorization on the speed of model training and testing. Besides, a unified CNN framework for both high-level and low-level vision tasks is provided, along with a vectorized Matlab implementation with state-of-the-art speed performance.

Some solvers, such as lingeling, use techniques The SAT Competitions, organized since 2002, have been the that cannot be expressed using resolution and cannot driving force of SAT solver development. The performance be expressed in the SAT Competition 2013 formats. of contemporary SAT solvers is incomparable to those of a One technique that cannot be expressed using resolution, decade ago. As a consequence, SAT solvers are used as the but is used in some top solvers, is bounded variable addition core search engine in many utilities, including tools for hardware (Manthey, Heule, and Biere 2013).

We present the SimSensei system, a fully automatic virtual agent that conducts interviews to assess indicators of psychological distress. We emphasize on the perception part of the system, a multimodal framework which captures and analyzes user state for both behavioral understanding and interactional purposes.

Graphical visualization has demonstrated enormous power in helping people to understand complexity in many branches of science. But, curiously, AI has been slow to pick up on the power of visualization. Alar is a visualization system intended to help people understand and control symbolic inference. Alar presents dynamically controllable node-and-arc graphs of concepts, and of assertions both supplied to the system and inferred. Alar is useful in quality assurance of knowledge bases (finding false, vague, or misleading statements; or missing assertions). It is also useful in tuning parameters of inference, especially how "liberal vs. conservative" the inference is Figure 1. An Alar visualization, centered on the assertion (trading off the desire to maximize the power of inference versus the risk of making incorrect inferences). We present "Orange is a food". Inferred assertions (green) use related a typical scenario of using Alar to debug a knowledge base.

The multi-agent path finding (MAPF) problem is a generalization of the single-agent path finding problem for k > 1 agents. It consists of a graph and a number of agents. Foreach agent, a unique start state and a unique goal state are given, the task is to find paths for all agents from their start states to their goal states, under the constraint that agents cannot collide during their movements. In many cases there is an additional goal of minimizing a cumulative cost function such as the sum of the time steps required for every agent to reach its goal. The goal of my research is providing new methods to solve MAPF optimally and provide theoretical understandings that will help choose the best solver given a problem instance.

We present an idea of using mathematicall modelling to guide a process of mining a set of patterns in an RDF graph and further exploiting these patterns to build expressive OWL class hierarchies.