Mining companies typically process all the material extracted from a mine site using processes which are extremely consumptive of energy and chemicals. Sorting the good material from the bad would effectively reduce required resources by leaving behind the bad material and only transporting and processing the good material. We use a relational influence diagram with an explicit utility model applied to the scenario in which an unknown number of rocks in unknown positions with unknown mineral compositions pass over 7 sensors toward 7 diverters on a high-throughput rock-sorting machine developed by MineSense Technologies Ltd. After receiving noisy sensor data, the system has 400 ms to decide whether to activate diverters which will divert the rocks into either a keep or discard bin. We learn the model offline and do online inference. Our result improves over the current state-of-the-art.

In this research-in-progress paper we present a new real world domain for studying the aggregation of different opinions: early stage architectural design of buildings. This is an important real world application, not only because building design and construction is one of the world's largest industries measured by global expenditures, but also because the early stage design decision making has a significant impact on the energy consumption of buildings. We present a mapping between the domain of architecture and engineering research and that of the agent models present in the literature. We study the importance of forming diverse teams when aggregating the opinions of different agents for architectural design, and also the effect of having agents optimizing for different factors of a multi-objective optimization design problem. We find that a diverse team of agents is able to provide a higher number of top ranked solutions for the early stage designer to choose from. Finally, we present the next steps for a deeper exploration of our questions.

Many scenarios require that robots work together as a team in order to effectively accomplish their tasks.  However, pre-coordinating these teams may not always be possible given the growing number of companies and research labs creating these robots.  Therefore, it is desirable for robots to be able to reason about ad hoc teamwork and adapt to new teammates on the fly.  This paper adopts an approach of learning policies to cooperate with past teammates and reusing these policies to quickly adapt to the new teammates.  This approach is applied to the complex domain of robot soccer in the form of half field offense in the RoboCup simulated 2D league.  This paper represents a preliminary investigation into this domain and presents a promising approach for tackling this problem.

As smart phones, tablets and similar computing devices become an integral part of our lives, we increasingly watch various types of streaming visual data on the display of those devices, especially as cloud video services become ubiquitous. One challenge is to present videos on diverse devices with an acceptable quality. Video retargeting is the key technology in video adaptation of cloud based video streaming. The most important challenge of video retargeting is to retain the shape of important objects, while ensuring temporal smoothness and coherence. We propose in this paper a new approach that adopts to the content of the video. We describe a cropping video retargeting method that ensures temporal coherence while enforcing spatial constraints by a saliency method. The average motion dynamics is calculated for each frame with optical flow and merged with the information of the user attention model for a given video. The resulting information is used to estimate a cropping window size. The output is a video that preserves important actions and the important parts of the scene. The results are promising in respect to overcoming the temporal and spatial challenges of video retargeting.

We reconceive editing for continuity and discontinuity, suggest how computers could be used to assist in such editing, and illustrate the artistic use of such editing in films and games.

We have proposed a novel data mining method called co-occurring cluster mining (CCM) for mining patterns from a sequence of multidimensional event data. The CCM first generates cluster candidates and then test the candidates based on clustering in the data space as well as co-occurrence degree in the event sequence. In searching appropriate clusters associated with co-occurrence, the search space is reduced by obtaining a dendrogram from a hierarchical clustering as the clustering procedure. In this paper, we show the potential of CCM with following two applications: (1) damage patterns in fuel cell and (2) earthquake occurrence patterns. In the fuel cell application, given a sequence of acoustic emission events, which comprise of waveform signal data of damages to a fuel cell, the mechanical interactions between components of the fuel cell are inferred from the mined co-occurrence patterns. Similarly, in the application of earthquakes, the interactions between distant earthquakes are extracted as co-occurrence patterns from a hypocenter catalog.

Opponent modeling is an essential approach for building competitive computer agents in imperfect information games. This paper presents a novel approach to develop opponent modeling techniques. The approach applies neural networks which are separately trained on different dataset to build K- model clustering opponent models. Kullback- Leibler (KL) divergence is used to exploit a safety mode on opponent modeling. Given a parameter d that controls the max divergence between a model’s centre point and the units belong to it, the approach is proved to provide a lower bound of expected payoff which is above the minimax payoff for correctly clustered players. Even for the players that are incorrectly clustered, the lower bound can also be unlimited approximated with sufficient history data. In our experiments, agent with the novel model shows an improved classification efficiency of opponent modeling comparing with relative researches. And also, the new agent performs better when playing against poker agent HITSZ_CS_13 which participate Annual Computer Poker Competition of 2013.

Online search in games has always been a core interest of artificial intelligence. Advances made in search for perfect information games (such as Chess, Checkers, Go, and Backgammon) have led to AI capable of defeating the world's top human experts. Search in imperfect information games (such as Poker, Bridge, and Skat) is significantly more challenging due to the complexities introduced by hidden information. In this paper, we present Online Outcome Sampling (OOS), the first imperfect information search algorithm that is guaranteed to converge to an equilibrium strategy in two-player zero-sum games. We show that OOS avoids common problems encountered by existing search algorithms and we experimentally evaluate its convergence rate and practical performance against benchmark strategies in Liar's Dice and a variant of Goofspiel. We show that unlike with Information Set Monte Carlo Tree Search (ISMCTS) the exploitability of the strategies produced by OOS decreases as the amount of search time increases. In practice, OOS performs as well as ISMCTS in head-to-head play while producing strategies with lower exploitability given the same search time.

Self-play reinforcement learning has proved to be successful in many perfect information two-player games. However, research carrying over its theoretical guarantees and practical success to games of imperfect information has been lacking. In this paper, we evaluate self-play Monte-Carlo Tree Search in limit Texas Hold'em and Kuhn poker. We introduce a variant of the established UCB algorithm and provide first empirical results demonstrating its ability to find approximate Nash equilibria.

Assistive technologies have the potential to enhance the quality of life of citizens. Most especially of interest are those cases where a person is affected by some physical or cognitive impairment. Whilst most work in this area have been focused on assisting people indoors to support their independence, the POSEIDON project is focused on empowering citizens with Down's Syndrome to support their independence outdoors. This paper explains the POSEIDON module which we are in the process of developing to make the system context-aware, reactive and adaptive.