Propositional Satisfiability (SAT) is fundamental in solving many application problems in Artificial Intelligence and Computer Science. Remarkable improvements in the efficiency of SAT solvers have beenmade over the last decade. Such improvements encourage researchers to solve constraint satisfaction problems by encoding them into SAT (i.e. ``SAT encoding''). Balanced Incomplete Block Design (BIBD) is one of the most typical block designs. BIBDs have been applied in several fields such as design experiments, coding theory, and cryptography. In this paper, we consider the problem of generating BIBDs by SAT encoding. We present a new SAT encoding that is an enhancement of order encoding with the idea of binary tree. It is designed to reduce the number of clauses required for cardinality constraints, compared with order encoding. In our experiments, our encoding was able to give a greater number of solutions than order encoding and state-of-the-art constraint solvers Mistral and choco.
In the field of the communication robots, many recent studies have focused on dialogue communication robots. This paper especially focused on the supporting robot for the conversation between humans. To help conversation between humans, we believe that the robots should have two abilities: estimate dialogue moods and behave suitably. In this paper, we propose dialogue moods estimation model. This paper, as the first step, focused on dialogues between two persons and construct estimation model for the dialogue moods observed by the third party. Because we believed that the dialogue moods are influenced by utterance time, which is extracted easily, the utterance intervals features are used to estimate the dialogue moods, for example, both solitary utterance intervals of leading speakers and following speakers, simultaneous utterance intervals, and silent intervals between two speakers. Using these utterance intervals features, we constructed the estimation model for dialogue moods by using Tree-Augmented Naive Bayes. Through the estimation experiment, we confirmed the availability of the estimation model for dialogue moods, in particular ``excitement,'' ``seriousness,'' and ``closeness,'' and the effective utterance intervals features for estimating dialogue moods. From the experimental results, it is suggested that the proposed model is effective for estimating dialogue moods.
Set Expansion is a method for finding similar sets of items from a seed set. It is useful on examining a large set of items to find hidden relationships among the items. In this paper, we propose a method that can reduce the number of calculations needed on executing Bayesian Sets, which is one of the most popular set expansion algorithms. The key point of our method lies in the use of Zero-suppressed Binary Decision Diagrams (ZDD) to express a binary value sparse matrix in a compressed form and executing needed calculations directly on constructed ZDD. We show a method for expressing a binary value matrix with ZDD, and also show some techniques for reducing the size of ZDD. We confirm the effectiveness of our method with experiments on both synthesis and real data.
This paper presents a new trial approach to early detection of dementia in the elderly with the use of functional brain imaging during cognitive tests. We have developed a non-invasive screening system of the elderly with cognitive impairment. In addition of our previous research of speech-prosody based data-mining approach, we had started the measurement of functional brain imaging for patient having a cognitive test by using functional near-infrared spectroscopy (fNIRS). We had collected 42 CHs fNIRS signals on frontal and right and left temporal areas from 50 elderly participants (18 males and 32 females between ages of 64 to 92) during cognitive tests in a specialized medical institute. We propose a Bayesian classifier, which can discriminate among elderly individuals with three clinical groups: normal cognitive abilities (NC), patients with mild cognitive impairment (MCI), and Alzheimer's disease (AD). The Bayesian classifier has two phases on the assumption of screening process, that firstly checks whether a suspicion of the cognitive impairment (CI) or not (NC) from given fNIRS signals; if any, and then secondly judges the degree of the impairment: cognitive impairment (MCI) or Alzheimer's disease (AD). This paper also reports the examination of the detection performance by cross-validation, and discusses the effectiveness of this study for early detection of cognitive impairment in elderly subjects. Consequently, empirical results that both the accuracy rate of AD and the predictive value of NC are equal to or more than 90\%. This suggests that proposed approach is adequate practical to screen the elderly with cognitive impairment.
I present a new text mining approach combined with network analysis to quantify the distribution patterns of the associated geographical names on a transportation network. extract geographical names from user's search queries recorded in search engine query logs and compare the similarities of any pair of geographical names using Jaccard coefficient. I found that a set of associated geographical names for each geographical name shows a specific spatial distribution pattern on transportation network and define a measure for quantifying such characteristics. Furthermore, I discuss its characteristics and application for information navigation.
An increase in short-term blood pressure (BP) variability is a characteristic feature in the elderly. It makes the management of hemodynamics more difficult, because it is frequently seen disturbed baro-reflex function and increased arterial stiffness, leading to isolated systolic hypertension. Large BP variability aggravates hypertensive target organ damage and is an independent risk factor for the cardiovascular (CV) events in elderly hypertensive patients. Therefore, appropriate control in BP is indispensable to manage lifestyle-related diseases and to prevent subsequent CV events. In addition, accumulating recent reports show that excessive BP variability is also associated with a decline in cognitive function and fall in the elderly. In the clinical settings, we usually evaluate their health condition, mainly with single point BP measurement using cuff inflation. However, unfortunately we are not able to find the close changes in BP by the traditional way. Here, we can show our advantageous approach of continuous BP monitoring using newly developing device `wearable BP sensing' without a cuff stress in the elderly. The new device could reflect systolic BP and its detailed changes, in consistent with cuff-based BP measurement. Our new challenge suggests new possibility of its clinical application with high accuracy.
Recentry, Multi-objective Genetic Algorithm, which is the application of Genetic Algorithm to Multi-objective Optimization Problems is focused on in the engineering design field. In this field, the analysis of design variables in the acquired Pareto solutions, which gives the designers useful knowledge in the applied problem, is important as well as the acquisition of advanced solutions. This paper proposes a new visualization method using Isomap which visualizes the geometric distances of solutions in the design variable space considering their distances in the objective space. The proposed method enables a user to analyze the design variables of the acquired solutions considering their relationship in the objective space. This paper applies the proposed method to the conceptual design optimization problem of hybrid rocket engine and studies the effectiveness of the proposed method.
Dungean semantics is known as the most standard argumentation-theoretic semantics that has the ability to evaluate various kinds of nonmonotonic consequences. The fact can be seen as a theoretical basis of adequacy for evaluating consequences of theoretical argumentation, i.e., argument about what to believe, that should be analyzed in terms of truth. It, however, does not give any theoretical basis of adequacy for evaluating consequences of practical argumentation, i.e., argument about what to do, that should be analyzed in terms of goodness. In order to address adequacy for evaluating consequences of practical argumentation formally, we propose practical argumentation semantics defined on practical argumentation frameworks. A practical argumentation framework has a defeat function specifying each agent's defeat relation caused by the agent's subjective desires, aims or values in conflicting arguments. Practical argumentation semantics is constructed on the notion of acceptability, which states that a set of agents accepts an argument if every argument defeating it is defeated by some argument under an agent in the set. We analyze properties of complete, preferred, grounded, and stable extensions defined for each set of agents. The correctness of our theory is shown by the fact that the practical argumentation semantics is a generalization of both Dungean semantics and Pareto optimality, i.e., a fundamental criterion for social welfare. Finally, we stratify practical argumentation frameworks on the basis of the theoretical facts and analyze properties of the frameworks.
We designed and practiced a cognitive science class for graduate students. In the class, the participants were required to build three cognitive models: a bug model, a trace model, and an individual model. In the construction of the bug model, the participants learn to construct a cognitive model by monitoring their mental processing. The participants confirmed that the trace model can explain human normative behavior; and also understood that the individual model can explain various patterns of human behavior that are generated by different problem solving strategies. The post questionnaire analysis shows that the participants successfully understood various aspects of advantages of the mode-based approach in cognitive science and important features of human cognitive processing.
We conducted an experimental investigation on human adaptation to change in an agent's strategy through a competitive two-player game. Modeling the process of human adaptation to agents is important for designing intelligent interface agents and adaptive user interfaces that learn a user's preferences and behavior strategy. However, few studies on human adaptation to such an agent have been done. We propose a human adaptation model for a two-player game. We prepared an on-line experimental system in which a participant and an agent play a repeated penny-matching game with a bonus round. We then conducted experiments in which different opponent agents (human or robot) change their strategy during the game. The experimental results indicated that, as expected, there is an adaptation phase when a human is confronted with a change in the opponent agent's strategy, and adaptation is faster when a human is competing with robot than with another human.
In this paper, interactive environment for learning by problem posing for reverse-thinking problems and its practical use are described. We have already developed several interactive learning environments for learning by problem-posing. In this research, we have paid a special attention to ``reverse-thinking problems'' in arithmetic word problems that can be solved either by addition or subtraction. In the reverse-thinking problems, since ``story operation structure'' and ``calculation operation structure'' are different, they require learners to comprehend the relations between problems and solutions more than ``forward thinking problems'' where ``story operation structure'' and ``calculation operation structure'' are the same ones. Based on a learning environment for posing the forward thinking problems developed previously, we have expanded it for reverse thinking problems. This learning environment has been used in a class of fourth grade at an elementary school for seven lesson times. We have also reported the results of this practical use.
Future robots/agents will perform situated behaviors for each user. Flexible behavioral learning is required for coping with diverse and unexpected users' situations. Unexpected situations are usually not tractable for machine learning systems that are designed for pre-defined problems. In order to realize such a flexible learning system, we were trying to create a learning model that can function in several different kinds of state transitions without specific adjustments for each transition as a first step. We constructed a modular neural network model based on reinforcement learning. We expected that combining a modular architecture with neural networks could accelerate the learning speed of neural networks. The inputs of our neural network model always include not only observed states but also memory information for any transition. In pure Markov decision processes, memory information is not necessary, rather it can lead to lower performance. On the other hand, partially observable conditions require memory information to select proper actions. We demonstrated that the new learning model could actually learn those multiple kinds of state transitions with the same architectures and parameters, and without pre-designed models of environments. This paper describes the performances of constructed models using probabilistically fluctuated Markov decision processes including partially observable conditions. In the test transitions, the observed state probabilistically fluctuated. The new learning model could function in those complex transitions. In addition, the learning speeds of our model are comparable to a reinforcement learning algorithm implemented with a pre-defined and optimized table-representation of states.