Understanding the mutual preferences between potential dating partners is core to the success of modern web-scale personalized recommendation systems that power online dating platforms. In contrast to classical user-item recommendation systems which model the unidirectional preferences of users to items, understanding the bidirectional preferences between people in a reciprocal recommendation system is more complex and challenging given the dynamic nature of interactions. In this paper, we describe a reciprocal recommendation system we built for one of the leading online dating applications in Japan. We also discuss the lessons learnt from designing, developing and deploying the reciprocal recommendation system in production.

This paper proposes a method for predicting human posture by thermal array sensors. In our previous work, we studied a system for estimating human posture by the sensors. The system successfully estimated human posture for elderly people living in a nursing home. As a further step to our previous work, this paper presents a human posture prediction method using Bayesian inference with the sensors. Our method predicts that posture changes at the next sample with Bayesian inference which employs characteristics of a past posture. In the experiment, we applied the results of our previous work and synthesized data of assumed daily movements. In conclusion, we obtained the basis of an effective posture prediction method using Bayesian inference and a thermal array sensor network.

This paper describes a human posture estimation system by using two thermal array sensors. In the system, the sensors are attached to ceiling and wall of a room, and acquire temperature distributions by 16 × 16 elements. The temperature distributions represent a state of temperature in the room, and they are analyzed to estimate human posture. Human posture is estimated by time-series posture transition diagram and the sum of temperature. In our experiment, we measured the temperature distributions in a room modeled as private room in a nursing home. As the results, the system successfully estimated human posture.

Support Vector Machines are a family of algorithms for the analysis of data based on convex Quadratic Programming. We derive randomized algorithms for training SVMs, based on a variation of Random Sampling Techniques; these have been successfully used for similar problems. We formally prove an upper bound on the expected running time which is quasilinear with respect to the number of data points and polynomial with respect to the other parameters, i.e., the number of attributes and the inverse of a chosen soft margin parameter. [This is the combined journal version of the conference papers (Balcázar, J.L. et al. in Proceedings of 12th International Conference on Algorithmic Learning Theory (ALT’01), pp. 119–134, [ 2001 ]; Balcázar, J.L. et al. in Proceedings of First IEEE International Conference on Data Mining (ICDM’01), pp. 43–50, [ 2001 ]; and Balcázar, J.L. et al. in Proceedings of SIAM Workshop in Discrete Mathematics and Data Mining, pp. 19–29, [ 2002 ]).]

We developed rescue robots and tools that are driven by hydraulic actuators. They are small-sized, have simple mechanisms, and are suitable for heavy-duty rescue operations. This report shows the results of field tests carried out to verify the possibility of the two rescue robots we developed; we named them the Jack Robot and the Cutter Robot. We classified disaster situations into three categories: collapsed house cases, collapsed building rubble cases, and traffic accidents. We used the two robots in test situations that simulated these three disaster scenarios. As part of the experimental results, we found that there is a great potential for the use of these high-powered small rescue robots in disaster areas.

In this paper, we propose a new type of haptic interface, named active link mechanism. This device realizes physical man-machine interaction (PMI) between machines and persons. Two prototypes were developed to demonstrate the potential of the active link mechanisms. Developed interface devices are an active tetrahedron and an active icosahedron. Nine-DOF micro spherical joints and pressure control pneumatic cylinders were developed to realize the active tetrahedron, while fifteen-DOF micro spherical joints and intelligent pneumatic cylinders were developed for the active icosahedron. The tetrahedron successfully realizes "virtual touch"; the operators feel actions, forces, and shapes of the virtual objects in PC and also move and deform them. Real time PMI is realized by building the developed devices into MSC.Visual-Nastran4D. MSC.VisuaI-Nastran4D is a mechanism analysis software, which can make motion analysis in real time. The active icosahedron also realized dynamic interaction with virtual objects in PC, showing the potential of the devices as a haptic interface.