

External AwardsLOIS Best Paper AwardWinners: Takeshi Kurashima^{*1}, Tomoharu Iwata^{*2}, Go Irie^{*1}, and Ko Fujimura^{*1} *1 NTT Cyber Solutions Laboratories *2 NTT Communication Science Laboratories Date: May 19, 2011 Organization: IEICE Life Intelligence and Office Information System (LOIS) Technical Group For ¡ÈTravel Route Recommendation Using Geotags on Photo Sharing Service¡É. Papers Published in Technical Journals and Conference ProceedingsSingleelectron Stochastic Resonance Using Si Nanowire TransistorsK. Nishiguchi and A. Fujiwara Proc. of the International Symposium on Nanoscale Transport and Technology (ISNTT2011), p. 133, Atsugi, Kanagawa, Japan. We demonstrate stochastic resonance (SR) with single electrons (SEs) using nanoscale metaloxidesemiconductor fieldeffect transistors (MOSFETs). The input signal applied to a MOSFET modulates SE transport in an average manner based on nonlinear characteristics. On the other hand, an individual SE goes through the MOSFET in a completely random manner, which corresponds to shot noise. SEs transferred to a storage node are counted precisely by the other MOSFET and used as an output signal. The correlation between the input and output signals is improved by taking advantage of extrinsic noise as well as the intrinsic shot noise composed of SEs. It is confirmed that the shotnoiseassisted SR allows fast operation with a simple system. Pattern perception utilizing SR is also demonstrated. A Silicon Nanowire Ionsensitive Fieldeffect Transistor with Elementary Charge SensitivityN. Clément, K. Nishiguchi, J. F. Dufreche, D. Guerin, A. Fujiwara, and D. Vuillaume Appl. Phys. Lett., Vol. 98, No. 1, p. 014104, 2011. We investigate the mechanisms responsible for the lowfrequency noise in liquidgated nanoscale silicon nanowire fieldeffect transistors (SiNWFETs) and show that the chargenoise level is lower than the elementary charge. Our measurements also show that the ionic strength of the surrounding electrolyte has a minimal effect on the overall noise. Dielectric polarization noise seems to be at the origin of the 1/f noise in our devices. The estimated spectral density of charge noise Sq = 1.6¡ß10^{¡Ý2} e/Hz^{1/2} at 10 Hz opens the door to metrological studies with these SiNWFETs for the electrical detection of a small number of molecules. Evaluation of a Gate Capacitance in the SubaF Range for a Chemical Fieldeffect Transistor with a Silicon Nanowire ChannelN. Clément, K. Nishiguchi, A. Fujiwara, and D. Vuillaume IEEE Trans. Nanotechnology, Vol. PP, No. 99, pp. 1–8, 2011. An evaluation of the gate capacitance of a fieldeffect transistor (FET) whose channel length and width are several tens of nanometers is a key point for sensors applications. However, experimental and precise evaluation of capacitance in the attofarad range or less has been extremely difficult. Here, we report an extraction of the capacitance down to 0.55 aF for a silicon FET with a nanoscale wire channel whose width and length are 15 and 50 nm, respectively. The extraction can be achieved by using a combination of four kinds of measurements: current characteristics modulated by double gates, randomtelegraphsignal noise induced by the trapping and detrapping of a single electron, dielectric polarization noise, and current characteristics showing Coulomb blockade at low temperature. The extraction of such a small gatecapacitance enables us to evaluate electron mobility in a nanoscale wire using a classical model of the current characteristics of a FET. A Theater for Viewing and Editing Multisensory ContentK. Hirota, S. Ebisawa, T. Amemiya, and Y. Ikei IEEE International Symposium on VR Innovation (ISVRI2011), pp. 239–244, Singapore. This research has been carried out as part of a multisensory theater project that aims at establishing technology to integrate a range of sensations such as visual, audio, force, tactile, vestibular, and odor into passive and interactive media content and communications. This paper describes the approaches that are being examined and the current status of the project. As a platform for the experiments, a prototype version of a multisensory theater has been implemented. The theater is equipped with devices that present wind and olfactory sensations. The sensation of wind is generated by both computercontrolled fans and air nozzles connected to a source of compressed air, and the olfactory sensation is presented by emitting odorants into the air. To facilitate the creation of content, a framework for editing multisensory information was constructed, in which all of the devices were connected to and controlled by a sequencer based on a MIDI interface. Standardization Trend of Recent Wireless LANsY. Kojima ARIB Bulletin, the Association of Radio Industries and Businesses (ARIB), No. 73, 2011 (in Japanese). This article explains the trend of recent wireless LAN standardization for IEEE802.11 and WiFi. Training Conditional Random Fields Based on Segmentwise Maximum Figureofmerit FunctionsJ. Suzuki and H. Isozaki Journal of IEICE, the Institute of Electronics, Information and Communication Engineers (IEICE), Vol. J94D, No. 5, pp. 908–918, 2011 (in Japanese). This paper proposes a framework for training conditional random fields (CRFs) to maximize a figureofmerit function for task evaluation. Specifically focusing on sequence segmentation tasks such as chunking and named entity recognition tasks in the natural language processing field, this paper introduces an objective function that emulates the segmentwise Fscore, which is generally used as an evaluation measure for these tasks. Our experiments show that the segmentwise Fscore optimization performs better than standard ML (maximum likelihood) and MAP (maximum a posteriori) training of CRF since it reduces the mismatch between the objective function of CRF training and the task evaluation measure. Singleelectron Counting Statistics of Shot Noise in Nanowire Si Metaloxidesemiconductor Fieldeffect transistorsK. Nishiguchi, Y. Ono, and A. Fujiwara Appl. Phys. Lett., Vol. 98, No. 19, p. 193502, 2011. Shot noise in the transport of single electrons in a Si metaloxidesemiconductor fieldeffect transistor is monitored by realtime measurement with a highchargesensitivity electrometer. In the current range between zepto and attoamperes, the current characteristics are found to be divided into two regimes: a temperatureindependent regime in the lower current range and a temperaturedependent one in the higher current range. A timedomain analysis reveals that, for both regimes, the singleelectron transport obeys a pure Poisson process with the Fano factor being nearly unity, while the shot noise is suppressed with reduced Fano factors around the boundary. High Output Power (~400 µW) Oscillators at around 550 GHz Using Resonant Tunneling Diodes with Graded Emitter and Thin BarriersM. Shiraishi, H. Shibayama, K. Ishigaki, S. Suzuki, M. Asada, H. Sugiyama, and H. Yokoyama. Appl. Phys. Express, Vol. 4, No. 6, p. 064101, 2011. We report resonant tunneling diode (RTD) oscillators with a high output power of around 400 µW at frequencies of 530–590 GHz. RTDs with a graded emitter and thin barriers were employed to obtain large negative differential conductance at high frequencies for high output power. An optimized structure of offset slot antennas was also used to maximize the radiation conductance. The highest output power obtained in this study was 420 µW at 548 GHz for an RTD with a peak current density of 24 mA/µm^{2}; the RTD was placed 58 µm away from the center of a 130µmlong slot antenna. Asymptotic Local Hypothesis Testing between a Pure Bipartite State and the Completely Mixed StateM. Owari and M. Hayashi arXiv, Cornell University Library, Vol. 1105, No. 3789, pp. 1–14, 2011. In this paper, we treat asymptotic hypothesis testing between an arbitrary known bipartite pure state and the completely mixed state under oneway local quantum operations assisted by classical communication (LOCC), twoway LOCC, and separable positive operator valued measures (POVMs). As a result, we derive analytical formulas for Stein¡Çs lemma type of optimal error exponents under all oneway LOCC, twoway LOCC, and separable POVMs; the Chernoff bounds under oneway LOCC POVMs and separable POVMs; and the Hoeffding bounds under oneway LOCC POVMs without any restrictions on a parameter and under separable POVMs in a restricted region of a parameter. We also numerically calculate the Chernoff and Hoeffding bounds for a class of threestep LOCC protocols in lowdimensional systems and show that these bounds not only outperform the bounds for oneway LOCC POVMs, but almost approximate the bounds for separable POVMs in the region of a parameter where the analytical bounds for separable POVMs are derived. Nonstationary Noise Estimation Method Based on Biasresidual Component Decomposition for Robust Speech RecognitionM. Fujimoto, S. Watanabe, and T. Nakatani Proc. of the 36th International Conference on Acoustics, Speech and Signal Processing (ICASSP2011), IEEE, pp. 4816–4819, Prague, Czech Republic. This paper addresses a noise suppression problem, namely the estimation of nonstationary noise sequences. In this problem, we assume that nonstationary noise can be decomposed into stationary and nonstationary components. These components are described respectively as the bias factor and the residual signal between the bias component and noise in each frame. This decomposition clarifies the role of each component, thus enabling us to apply a suitable parameter estimation technique to each component. In this paper, the bias component is estimated by the EM (expectation maximization) algorithm with the entire observed signal sequence. On the other hand, the residual component is sequentially estimated by multiplying the extended Kalman filter with the EM algorithm. In the evaluation results, we confirmed that the proposed method improved speech recognition accuracy compared with noise estimation methods without component decomposition. 