Dynamics of learning near singularities in layered networks

We explicitly analyze the trajectories of learning near singularities in hierarchical networks, such as multilayer perceptrons and radial basis function networks, which include permutation symmetry of hidden nodes, and show their general properties. Such symmetry induces singularities in their parameter space, where the Fisher information matrix degenerates and odd learning behaviors, especially the existence of plateaus in gradient descent learning, arise due to the geometric structure of singularity. We plot dynamic vector fields to demonstrate the universal trajectories of learning near singularities. The singularity induces two types of plateaus, the on-singularity plateau and the near-singularity plateau, depending on the stability of the singularity and the initial parameters of learning. The results presented in this letter are universally applicable to a wide class of hierarchical models. Detailed stability analysis of the dynamics of learning in radial basis function networks and multilayer perceptrons will be presented in separate work.     

Negative emotion recognition using multimodal physiological signals for advanced driver assistance systems

Artificial Life and Robotics - Recent advanced driver assistance systems’ (ADASs) control cars to avoid accidents, but few of them consider driver’s comfort. To realize comfortable...     

Intercalation of n-alkylamines into α-tin(IV) bis(hydrogenphosphate)

The intercalation reaction of n-alkylamines from methyl- to hexadecylamine into -tin(IV) bis(hydrogenphosphate) has been investigated. The reaction was conducted by mixing the phosphate and solutions of the amines dissolved in water or benzene. The amine/ phosphate molar ratios in the reaction mixtures were 0.43, 0.85, 1.71 and 3.41. The nitrogen/phosphorus atomic ratio in the intercalation compound was not changed by the amine/phosphate molar ratio. The interlayer distances of the intercalates obtained in the reaction mixtures with molar ratios of 3.41 and 1.71 increased with an increase in the number of carbon atoms in the alkyl chain of the amine. The slope of the straight line obtained in plot of interlayer distance versus number of carbon atoms in the alkyl chain of the amine indicates that the amine molecules form a double layer in the interlayer space of the phosphate and are inclined at an approximate angle of 67.7° to the phosphate layer. The interlayer distances of the intercalates of butyl- to nonylamines obtained at the molar ratio of 0.85 are smaller than those of the corresponding compounds from the molar ratios of 3.41 and 1.71, while those of decyl- to hexadecylamine intercalates lie on the previously discussed line. This behaviour is interpreted by assuming kink formation in the short alkyl chains between the phosphate layers.     

Digital RAC with a disturbance observer for underwater vehicle-manipulator systems

Most control methods of underwater vehiclemanipulator systems (UVMS) are based on the computed torque method that is used for underwater robotic vehicles. We have proposed a resolved acceleration control (RAC) method for UVMS. In this article, we propose a disturbance compensation control method for both vehicle and manipulator based on the RAC method. Experimental results using an underwater robot with a vertical planar 2-link manipulator show that the proposed control method has good control performance.     

Robust comprehension of natural language instructions by a domestic service robot

We present a method through which domestic service robots can comprehend natural language instructions. For each action type, a variety of natural language expressions can be used, for example, the instruction, ‘Go to the kitchen’ can also be expressed as ‘Move to the kitchen.’ We are of the view that natural language instructions are intuitive and, therefore, constitute one of the most user-friendly robot instruction methods. In this paper, we propose a method that enables robots to comprehend instructions spoken by a human user in his/her natural language. The proposed method combines action-type classification, which is based on a support vector machine, and slot extraction, which is based on conditional random fields, both of which are required in order for a robot to execute an action. Further, by considering the co-occurrence relationship between the action type and the slots along with the speech recognition score, the proposed method can avoid degradation of the robot’s comprehension accuracy in noisy environments, where inaccurate speech recognition can be problematic. We conducted experiments using a Japanese instruction data-set collected using a questionnaire-based survey. Experimental results show that the robot’s comprehension accuracy is higher in a noisy environment using our method than when using a baseline method with only a 1-best speech recognition result.     

Prediction of carbon dioxide separation from gas mixtures in petroleum industries using the Levenberg–Marquardt algorithm

In this study, two mathematical models for hydrate formation process to separate carbon dioxide by a combination of two different kinds of organic and surfactant promoters are presented. Promoters such as sodium dodecyl sulfate, sodium dodecyl benzene sulfonate, and dodecyl trimethyl ammonium chloride as surfactant promoters; also, tetrahydrofuran, cyclopentane, 1,3-dioxolane, and 2-methyl tetrahydrofuran as organic promoters have been used in recent years. The results showed that a combination of 3000 ppm of surfactant promoters and 4 wt% organic promoters had the highest separation rate of carbon dioxide and; consequently, the investigated models were based on this optimum condition. As a matter of fact, by using these simulations the hydrate formation behavior was predicted with high accuracy; moreover, conducting consuming experiments is not essential anymore. To sum up, in the present research both Vandermonde matrix model and Levenberg-Marquardt algorithm were applied to predict the hydrate formation behavior; in addition, their results were precisely considered and validated with experimental data.     

Intestinal absorption and biological effects of orally administered amorphous silica particles

Although amorphous silica nanoparticles are widely used in the production of food products (e.g., as anticaking agents), there is little information available about their absorption and biological effects after oral exposure. Here, we examined the in vitro intestinal absorption and in vivo biological effects in mice of orally administered amorphous silica particles with diameters of 70, 300, and 1,000 nm (nSP70, mSP300, and mSP1000, respectively) and of nSP70 that had been surface-modified with carboxyl or amine groups (nSP70-C and nSP70-N, respectively). Analysis of intestinal absorption by means of the everted gut sac method combined with an inductively coupled plasma optical emission spectrometer showed that the intestinal absorption of nSP70-C was significantly greater than that of nSP70. The absorption of nSP70-N tended to be greater than that of nSP70; however, the results were not statistically significant. Our results indicate that silica nanoparticles can be absorbed through the intestine and that particle diameter and surface properties are major determinants of the degree of absorption. We also examined the biological effects of the silica particles after 28-day oral exposure in mice. Hematological, histopathological, and biochemical analyses showed no significant differences between control mice and mice treated with the silica particles, suggesting that the silica nanoparticles evaluated in this study are safe for use in food production.     

Production of hyaluronic‐acid‐based cell‐enclosing microparticles and microcapsules via enzymatic reaction using a microfluidic system

This article describes the preparation of cell‐enclosing hyaluronic acid (HA) microparticles with solid core and microcapsules with liquid core through cell‐friendly horseradish peroxidase (HRP)‐catalyzed hydrogelation. The spherical vehicles were made from HA derivative possessing phenolic hydroxyl moieties (HA‐Ph) cross‐linkable through the enzymatic reaction by extruding cell‐suspending HA‐Ph aqueous solution containing HRP from a needle of 180 μm in inner diameter into the ambient coaxial flow of liquid paraffin containing H₂O₂ in a microtubule of 600 μm in diameter. By altering the flow rate of liquid paraffin, the diameters of gelatin and HA‐Ph microparticles were varied in the range of 120–220 μm and 100–300 μm, respectively. The viability of the enclosed human hepatoma HepG2 cells in the HA‐Ph microparticles of 180 μm in diameter was 94.2 ± 2.3%. The growth of the enclosed HepG2 cells was enhanced by decreasing the HRP concentration. The microcapsules of 200 μm in diameter were obtained by extruding HA‐Ph aqueous solution containing thermally liquefiable cell‐enclosing gelatin microparticles of 150 μm in diameter using the same microfluidic system. The enclosed cells grew and filled the cavity within 10 days. Spherical tissues covered with a heterogeneous cell layer were obtained by degrading the microcapsule membrane using hyaluronidase after covering the surface with a heterogeneous cell layer. © 2015 Wiley Periodicals, Inc. J. Appl. Polym. Sci. 2016 , 133, 43107.     

Epitaxial p-type SiC as a self-driven photocathode for water splitting

Solar-to-hydrogen conversion efficiencies of water-splitting photochathodes using epitaxially grown p-type 4H-, 6H- and 3C-SiC were estimated in a two-electrode system without applying any external bias. By using electrode materials with small oxygen overpotentials as counter electrodes, the photocurrent became comparable to that observed in a three-electrode system with a suitable bias. Estimated efficiencies seem to depend on the bandgap of the SiC polytypes. For the 3C-SiC, the obtained efficiency was 0.38%, which is so far the highest value reported for SiC. We confirmed that the hydrogen volumes estimated from the photocurrent were almost the same as actual volumes observed by gas chromatography.