Teaching robots about objects in their environment requires a multimodal correlation of images and linguistic descriptions to build complete feature and object models.  These models can be created manually by collecting images and related keywords and presenting the pairings to robots, but doing so is tedious and unnatural.  This work abstracts the problem of training robots to learn about the world around them by introducing I Spy , an interactive dialogue- and vision-based game in which players place objects in front of a humanoid robot and challenge it to guess which object they have in mind.  The robot gradually learns about the objects and the features which describe them through repeated games, by updating its knowledge with newly captured training images.  This paper details I Spy's learning and gaming processes, describes the approaches taken to extract information from multiple modalities both before and during gameplay, and finally discusses the results of a study designed to evaluate the game's model accuracy over time, its overall performance, and its appeal to human players.

In the context of soccer robotics, ball dribbling is a complex behavior where a robot player attempts to maneuver the ball in a very controlled way, while moving towards a desired target. To learn when and how to modify the robot’s velocity vector is a complex problem, hardly solvable in an effective way with methods based on identification of the system dynamics and/or kinematics and mathematical models. We propose a decentralized reinforcement learning strategy, where each component of the omnidirectional biped walk (𝑣𝑥,𝑣𝑦,𝑣𝜃) is learned in parallel with single-agents working in a multiagent task. Moreover, we propose an approach to accelerate the decentralized learning based on knowledge transfer from simple linear controllers. Obtained results are successful; with less human effort, and less required designer knowledge, the decentralized reinforcement learning scheme shows better performances than the current dribbling engine used by UChile Robotics Team in the SPL robot soccer competitions. The proposed decentralized rein- forcement learning scheme achieves asymptotic performance after 1500 episodes and can be accelerated up to 70% by using our approach to share actions.

Reinforcement learning enables a robot to learn behavior through trial-and-error. However, knowledge is usually built from scratch and learning may take a long time. Many approaches have been proposed to transfer the knowledge learned in one task and reuse it in another new similar task to speed up learning in the target task.A very effective knowledge to be transferred is an abstract policy, which generalizes the learned policies in source tasks to extend the domain of tasks that can reuse them.There are inductive and deductive methods to generate abstract policies.However, there is a lack of deeper analysis to assess not only the effectiveness of each type of policy, but also the way in which each policy is used to accelerate the learning in a new task.In this paper we propose two simple inductive methods and we use a deductive method to generate stochastic abstract policies from source tasks. We also propose two strategies to use the abstract policy during learning in a new task: the hard and the soft strategy. We make a comparative analysis between the three types of policies and the two strategies of use in a robotic navigation domain.We show that these techniques are effective in improving the agent learning performance, especially during the early stages of the learning process, when the agent is completely unaware of the new task.

Cities and their employment catchment areas are focus points of economic activity, transportation, and social interactions. The need for land use and transport inte- grated modelling (LUTI modelling) as a decision aid tool in urban planning, has become apparent. Instanti- ating such models on cities, requires a substantial data collection, model structuring and parameter estimation effort; for conciseness, the latter is referred to here as calibration. This work is a partial effort towards the integrated calibration of LUTI models. It considers one of the most widely used LUTI models and softwares, Tranus. The usual calibration approach for Tranus is briefly reviewed. It is then reformulated as an optimisa- tion problem, in order to make it amenable to the sys- tematic incorporation of constraints on parameters and additional data and to form a clear basis for future fully integrated calibration. The problem at hand concerns a dynamic system; an approach is shown how to “elimi- nate” parts of the dynamics in order to ease the param- eter optimisation. We also discuss how to validate cali- bration results and propose to use synthetic data gener- ated from real world problems in order to assess conver- gence properties and accuracy of calibration methods.

We aim to profile characteristics of areas of variant units across a district, city or a country. Studying attributes of areas can be very useful in several situations. In the past, research has focused mainly on studying specific char- acteristics of areas using a few selected attributes. In this paper we propose an alternative view on neighbourhood profiles. Instead of characterising a neighbourhood through a set of attributes such as those collected by the census, we propose use of a low-dimensional fea- ture representation, or embedding, created from one or more input sources. The purpose of the embeddings is having a generic representation for entities that can do well across several downstream tasks such as regression for attributes prediction.

This is especially the media as an event unfolds. This methodology consists of case in emergency situations, where the spread of a false rumour three main steps: (i) collection of (source) tweets posted during can have dangerous consequences. For instance, in a an emergency situation, sampling in such a way that situation where a hurricane is hitting a region, or a terrorist it is manageable for human assessment, while generating attack occurs in a city, access to accurate information is a good number of rumourous tweets from multiple stories, crucial for finding out how to stay safe and for maximising (ii) collection of conversations associated with each of the citizens' wellbeing. This is even more important in cases source tweets, which includes a set of replies discussing the where users tend to pass on false information more often source tweet, and (iii) collection of human annotations on than real facts, as occurred with Hurricane Sandy in 2012 the tweets sampled. We provide a definition of a rumour (Zubiaga and Ji 2014). Hence, identifying rumours within a which informs the annotation process. Our definition draws social media stream can be of great help for the development on definitions from different sources, including dictionaries of tools that prevent the spread of inaccurate information.

Soil organic carbon (SOC) plays a major role in the global carbon budget. It can act as a source or a sink of atmospheric carbon, thereby possibly influencing the course of climate change. Improving the tools that model the spatial distributions of SOC stocks at national scales is a priority, both for monitoring changes in SOC and as an input for global carbon cycles studies. In this paper, we compare and evaluate two recent and promising modelling approaches. First, we considered several increasingly complex boosted regression trees (BRT), a convenient and efficient multiple regression model from the statistical learning field. Further, we considered a robust geostatistical approach coupled to the BRT models. Testing the different approaches was performed on the dataset from the French Soil Monitoring Network, with a consistent cross-validation procedure. We showed that when a limited number of predictors were included in the BRT model, the standalone BRT predictions were significantly improved by robust geostatistical modelling of the residuals. However, when data for several SOC drivers were included, the standalone BRT model predictions were not significantly improved by geostatistical modelling. Therefore, in this latter situation, the BRT predictions might be considered adequate without the need for geostatistical modelling, provided that i) care is exercised in model fitting and validating, and ii) the dataset does not allow for modelling of local spatial autocorrelations, as is the case for many national systematic sampling schemes.

Event detection has been one of the most important research topics in social media analysis. Most of the traditional approaches detect events based on fixed temporal and spatial resolutions, while in reality events of different scales usually occur simultaneously, namely, they span different intervals in time and space. In this paper, we propose a novel approach towards multiscale event detection using social media data, which takes into account different temporal and spatial scales of events in the data. Specifically, we explore the properties of the wavelet transform, which is a well-developed multiscale transform in signal processing, to enable automatic handling of the interaction between temporal and spatial scales. We then propose a novel algorithm to compute a data similarity graph at appropriate scales and detect events of different scales simultaneously by a single graph-based clustering process. Furthermore, we present spatiotemporal statistical analysis of the noisy information present in the data stream, which allows us to define a novel term-filtering procedure for the proposed event detection algorithm and helps us study its behavior using simulated noisy data. Experimental results on both synthetically generated data and real world data collected from Twitter demonstrate the meaningfulness and effectiveness of the proposed approach. Our framework further extends to numerous application domains that involve multiscale and multiresolution data analysis.

Those of us involved In the creation of the Handbook of Artificial Intelligence, both writers and editors, have attempted to make the concepts, methods, tools, and main results of artificial Intelligence research accessible to a broad scientific and engineering audience. Currently, Al work Is familiar mainly to its practicing specialists and other interested computer scientists. Yet the field Is of growing interdisciplinary interest and practical Importance. With this book we are trying to build bridges that are easily crossed by engineers, scientists in other fields, and our own computer science colleagues. In the Handbook we Intend to cover the breadth and depth of Al, presenting general overviews of the scientific issues, as well as detailed discussions of particular to -hniques and Important Al systems.

There is little doubt that the decision to use rules to encode infectious disease knowledge in the nascent MYCIN system was largely influenced by our experience using similar techniques in DENDRAL. However, as mentioned in Chapter 1, we did experiment with a semantic network representation before turning to the production rule model. The impressive published examples of Carbonell's SCHOLAR system (Carbonell, 1970a; 1970b), with its ability to carry on a mixed-initiative dialogue regarding the geography of South America, seemed to us a useful model of the kind of rich interactive environment that would be needed for a system to advise physicians. Our disenchantment with a pure semantic network representation of the domain knowledge arose for several reasons as we began to work with Cohen and Axline, our collaborating experts. First, the knowledge of infectious disease therapy selection was ill-structured and, we found, difficult to represent using labeled arcs between nodes. Unlike South American geography, our domain did not have a clear-cut hierarchical organization, and we found it challenging to transfer a page or two from a medical textbook into a network of sufficient richness for our purposes. Of particular importance was our need for a strong inferential mechanism that would allow our system to reason about complex relationships among diverse concepts; there was no precedent for inferences on a semantic net that went beyond the direct, labeled relationships between nodes.1 Perhaps the greatest problem with a network representation, and the greatest appeal of production rules, was our gradually recognized need to deal with small chunks of domain knowledge in interacting with our expert collaborators.