Locally linear embedding (LLE) is a well-known method for nonlinear dimensionality reduction. The mathematical proof and experimental results presented in this paper show that the neighborhood sizes in LLE must be smaller than the dimensions of input data spaces, otherwise LLE would degenerate from a nonlinear method for dimensionality reduction into a linear method for dimensionality reduction. Furthermore, when the neighborhood sizes are larger than the dimensions of input data spaces, the solutions to LLE are not unique. In these cases, the addition of some regularization method is often proposed. The experimental results presented in this paper show that the regularization method is not robust. Too large or too small regularization parameters cannot unwrap S-curve. Although a moderate regularization parameters can unwrap S-curve, the relative distance in the input data will be distorted in unwrapping. Therefore, in order to make LLE play fully its advantage in nonlinear dimensionality reduction and avoid multiple solutions happening, the best way is to make sure that the neighborhood sizes are smaller than the dimensions of input data spaces.

Boosting over weak classifiers is widely used in pedestrian detection. As the number of weak classifiers is large, researchers always use a sampling method over weak classifiers before training. The sampling makes the boosting process harder to reach the fixed target. In this paper, we propose a partial derivative guidance for weak classifier mining method which can be used in conjunction with a boosting algorithm. Using weak classifier mining method makes the sampling less degraded in the performance. It has the same effect as testing more weak classifiers while using acceptable time. Experiments demonstrate that our algorithm can process quicker than [1] algorithm in both training and testing, without any performance decrease. The proposed algorithms is easily extending to any other boosting algorithms using a window-scanning style and HOG-like features.

We propose a novel method of supervised feature projection called class-distance-based discriminant analysis (CDDA), which is suitable for automatic age estimation (AAE) from facial images. Most methods of supervised feature projection, e.g., Fisher discriminant analysis (FDA) and local Fisher discriminant analysis (LFDA), focus on determining whether two samples belong to the same class (i.e., the same age in AAE) or not. Even if an estimated age is not consistent with the correct age in AAE systems, i.e., the AAE system induces error, smaller errors are better. To treat such characteristics in AAE, CDDA determines between-class separability according to the class distance (i.e., difference in ages); two samples with similar ages are imposed to be close and those with spaced ages are imposed to be far apart. Furthermore, we propose an extension of CDDA called local CDDA (LCDDA), which aims at handling multimodality in samples. Experimental results revealed that CDDA and LCDDA could extract more discriminative features than FDA and LFDA.

This paper introduces a random selection cooperation scheme that takes the Decode-and-Forward (DF) approach to solve the unfairness problem in selection cooperation. Compared to previous work which obtained fairness but introduced performance loss, the proposed scheme guarantees fairness without performance loss. Its essence is to randomly select from the relays that can ensure the successful communication between the source and the destination, rather than to select the best relay. Both a theoretical analysis and simulation results confirm that the proposed scheme could achieve fairness and introduce no performance loss. We also discuss the conditions under which the proposed scheme is practical to implement.

This paper proposes an active-RC filter that achieves a wide pseudo-continuous bandwidth-tuning range and a wide gain range with fine steps using a novel switched resistor architecture. A channel-selection filter with the proposed resistor bank is designed for a multi-mode mobile-TV receiver with the 6th order Chebyshev-I topology. The bandwidth, 0.5–6 MHz with 5% steps, supports multiple mobile-TV standards with sufficient margins for process and temperature variations. The filter also accomplishes a 30-dB variable gain range with 6-dB steps, and it relaxes the dynamic range requirement of a succeeding programmable gain amplifier. The power consumption of the filter, 3.4–5.0 mW, is adjustable according to the bandwidth and the signal level. The filter was fabricated with on-chip bandwidth-calibration circuitry in 0.18-µm CMOS and occupied 0.81 mm2.

Detecting distributed anomalies rapidly and accurately is critical for efficient backbone network management. In this letter, we propose a novel anomaly detection method that uses router connection relationships to detect distributed anomalies in the backbone Internet. The proposed method unveils the underlying relationships among abnormal traffic behavior through closed frequent graph mining, which makes the detection effective and scalable.

This correspondence contributes to some d-form functions and d-form sequences. A property of d-form functions is obtained firstly. Then we present a way to construct d-form sequences and extended d-form sequences with ideal autocorrelation. Based on our result, many sequences with ideal autocorrelation can be constructed by the corresponding difference-balanced d-form functions.

This work describes a 12-bit 100 MS/s 0.13 µm CMOS three-stage pipeline ADC with various circuit design techniques to reduce power and die area. Digitally controlled timing delay and gate-bootstrapping circuits improve the linearity and sampling time mismatch of the SHA-free input network composed of an MDAC and a FLASH ADC. A single two-stage switched op-amp is shared between adjacent MDACs without MOS series switches and memory effects by employing two separate NMOS input pairs based on slightly overlapped switching clocks. The interpolation, open-loop offset sampling, and two-step reference selection schemes for a back-end 6-bit flash ADC reduce both power consumption and chip area drastically compared to the conventional 6-bit flash ADCs. The prototype ADC in a 0.13 µm CMOS process demonstrates measured differential and integral non-linearities within 0.44LSB and 1.54LSB, respectively. The ADC shows a maximum SNDR and SFDR of 60.5 dB and 71.2 dB at 100 MS/s, respectively. The ADC with an active die area of 0.92 mm2 consumes 19 mW at 100 MS/s from a 1.0 V supply. The measured FOM is 0.22 pJ/conversion-step.

Since retina blood vessels (RBV) are a major factor in ophthalmological diagnosis, it is essential to detect RBV from a fundus image. In this letter, we proposed the detection method of RBV using a morphological analysis and support vector machine classification. The proposed RBV detection method consists of three strategies: pre-processing, features extraction and classification. In pre-processing, noises were reduced and RBV were enhanced by anisotropic diffusion filtering and illumination equalization. Features were extracted by using the image intensity and morphology of RBV. And a support vector machine (SVM) classification algorithm was used to detect RBV. The proposed RBV detection method was simulated and validated by using the DRIVE database. The averages of accuracy and TPR are 0.94 and 0.78, respectively. Moreover, by comparison, we confirmed that the proposed RBV detection method detected RBV better than the recent RBV detections methods.

A Boolean function is said to be correlation immune if its output leaks no information about its input values. Such functions have many applications in computer security practices including the construction of key stream generators from a set of shift registers. Finding methods for easy construction of correlation immune Boolean functions has been an active research area since the introduction of the notion by Siegenthaler. In this paper, we present several constructions of nonpalindromic correlation immune symmetric Boolean functions. Our methods involve finding binomial coefficient identities and obtaining new correlation immune functions from known correlation immune functions. We also consider the construction of higher order correlation immunity symmetric functions and propose a class of third order correlation immune symmetric functions on n variables, where n+1(≥ 9) is a perfect square.

This paper provides conceptual and experimental analysis of a new approach in the study of kanji, our “Learner's Visualization (LV) Approach”. In a previous study we found that the LV Approach assists beginning learners in significantly updating their personal kanji deconstruction visualization. Additionally, in another study our findings provided evidence that beginning learners also receive a significant impact in the ability to acquire vocabulary. In this study, our research problem examines how beginning and intermediate students use visualization to cognitively deconstruct (divide) kanji in different ways, and how this affects their learning progress. We analyze the cognitive differences in how kanji learners explore and deconstruct novel kanji while using the LV Approach and how these differences affect their learning process while using the LV Approach. During the learning experience, our LILES System (Learner's Introspective Latent Envisionment System), based on the LV Approach, guides learners to choose from a set of possible “kanji deconstruction layouts” (layouts showing different ways in which a given kanji can be divided). The system then assists learners in updating their “kanji deconstruction level” (the average number of parts they visualize within kanji according to their current abilities). Statistical analysis based on achieved performance was conducted. The analysis of our results proves that there are cognitive differences: beginners deconstruct kanji into more parts (“blocks”) than intermediate learners do, and while both improve their kanji deconstruction scores, there is a more significant change in “kanji deconstruction level” in beginners. However, it was also found that intermediate learners benefit more in “kanji retention score” compared with beginners. Suggestions for further research are provided.

In this paper, a novel receiver architecture is proposed for multiple-input-multiple-output (MIMO) systems; the proposed architecture helps achieve superior performance in multipath fading channels when the number of layered streams exceeds the number of receiving antennas. In this architecture, the concept of “virtual channel” is adopted to attain diversity gain even when successive detection is applied for reducing computational complexity, while orthogonal frequency division multiplexing (OFDM) is employed to combat multipath fading. Actually, successive detection is carried out in all possible virtual channels, and the virtual channel with the minimum error probability is detected with the assistance of the maximum a-posteriori (MAP) decoder in the architecture. In addition, soft input and soft output (SISO) iterative detection is introduced in the virtual channel estimation scheme. The performance of the proposed architecture is verified by computer simulations. This architecture can be implemented with lesser complexity than that in maximum likelihood detection (MLD), but the gain in the former case exceeds that in the latter by 4.5 dB at the BER of 10-3 for 42 MIMO-OFDM.

Silicon Mach-Zehnder interferometric (MZI) waveguide modulator incorporating the n-channel junction field-effect transistor (JFET) as a signal modulation unit was designed, fabricated, and analyzed. The proposed MZI with JFET was designed to operate based on the plasma dispersion effect in the infrared wavelength of 1550 nm. The three different modulation lengths (ML) of 500, 1000, and 2000 µm while keeping the overall MZI length constant at 1.5 cm were set as a general design rule for these 10 µm-wide MZIs under study. When the JFET was operated in an active mode by injecting approximately 50 mA current (Is) to achieve a π phase shift, the modulation efficiency of the device was measured to be η = π /(Is· L) 40π/A-mm. The temporal and frequency response measurements also demonstrate that the respectively rise and fall times measured using a high-speed photoreceiver were in the neighborhood of 8.5 and 7.5 µsec and the 3 dB roll-off frequency (f3 dB) measured was in the excess of 400 kHz.

The present paper introduces a new approach to the construction of a sequence set with a zero-correlation zone for both periodic and aperiodic correlation functions. The proposed sequences can be constructed from a pair of Hadamard matrices of orders n0 and n1. The constructed sequence set consists of n0 n1 ternary sequences, each of length n0(m+2)(n1+Δ), for a non-negative integer m and Δ ≥ 2. The zero-correlation zone of the proposed sequences is |τ| ≤ n0m+1-1, where τ is the phase shift. The proposed sequence set consists of n0 subsets, each with a member size n1. The correlation function of the sequences of a pair of different subsets, referred to as the inter-subset correlation function, has a zero-correlation zone with a width that is approximately Δ times that of the correlation function of sequences of the same subset (intra-subset correlation function). The inter-subset zero-correlation zone of the proposed sequences is |τ| ≤ Δn0m+1, where τ is the phase shift. The wide inter-subset zero-correlation enables performance improvement during application of the proposed sequence set.

A bidirectional amplification module has been proposed for use in IP-over-CWDM networks. The module is based on a bidirectional erbium-doped fiber amplifier. The loss compensation characteristics of the module obtained in a bidirectional IP transmission experiment confirmed that the losses of the optical node and the transmission fiber in the network can be compensated for effectively by the module making it possible to increase the number of nodes and the total fiber length of the network.

In this paper, a novel channel estimation method for time domain synchrotrons orthogonal frequency domain multiplexing (TDS-OFDM) based on training sequence cyclic reconstruction is proposed to eliminate residual inter-block interference (IBI); it estimates the channel impulse response (CIR) in an iterative manner. A simulation and analysis show that the proposed method can effectively perform the channel estimation over long-delay multipath channels with low complexity.

A stereo video coding scheme which is compatible with monoview-processor is presented in this paper. At the same time, this paper proposes an adaptive prediction structure which can make different prediction modes to be applied to different groups of picture (GOPs) according to temporal correlations and interview correlations to improve the coding efficiency. Moreover, the most advanced video coding standard H.264 is used conveniently for maximize the coding efficiency in this paper. Finally, the effectiveness of the proposed scheme is verified by extensive experimental results.

We propose an active noise control (ANC) system for reducing periodic noise generated in a high magnetic field such as noise generated from magnetic resonance imaging (MRI) devices (MR noise). The proposed ANC system utilizes optical microphones and piezoelectric loudspeakers, because specific acoustic equipment is required to overcome the high-field problem, and consists of a head-mounted structure to control noise near the user's ears and to compensate for the low output of the piezoelectric loudspeaker. Moreover, internal model control (IMC)-based feedback ANC is employed because the MR noise includes some periodic components and is predictable. Our experimental results demonstrate that the proposed ANC system (head-mounted structure) can significantly reduce MR noise by approximately 30 dB in a high field in an actual MRI room even if the imaging mode changes frequently.

The tag collection algorithm in ISO/IEC 18000-7 has difficulty in collecting data from massive numbers of active RFID tags in a timely manner, so it should be improved to allow successful application in a wide variety of industrial fields. We propose two novel methods, a reduced-message method to improve the performance of data-tag collection and an efficient-sleep method to improve the performance of ID-tag collection. The reduced-message method decreases the slot size for a tag response by reducing the response size from the tag and reduces the number of commands issued from the reader. The efficient-sleep method utilizes redundant empty slots within the frame period to transmit sleep commands to the tags collected previously. We evaluated the performance improvement of tag collection by the proposed methods experimentally using an active RFID reader and 60 tags that we prepared for this study. The experimental results showed that the reduced-message method and the efficient-sleep method decreased the average tag collection time by 16.7% for data-tag collection and 9.3% for ID-tag collection compared with the standard tag collection. We also developed a simulation model for the active RFID system, reflecting the capture effect in wireless communication, and performed simulations to evaluate the proposed methods with a massive number of tags. The simulation results with up to 300 tags confirmed that the proposed methods could improve the tag collection performance, confirming the experimental results, even with larger numbers of tags.

Power consumption has become an increasing concern in high performance microprocessor design. Especially, Instruction Cache (I-Cache) contributes a large portion of the total power consumption in a microprocessor, since it is a complex unit and is accessed very frequently. Several studies on low-power design have been presented for the power-efficient cache design. However, these techniques usually suffer from the restrictions in the traditional Instruction Fetch Unit (IFU) architectures where the fetch address needs to be sent to I-Cache once it is available. Therefore, work to reduce the power consumption is limited after the address generation and before starting an access. In this paper, we present a new power-aware IFU architecture, named Analysis Before Starting an Access (ABSA), which aims at maximizing the power efficiency of the low-power designs by eliminating the restrictions on those low-power designs of the traditional IFU. To achieve this goal, ABSA reorganizes the IFU pipeline and carefully assigns tasks for each stages so that sufficient time and information can be provided for the low-power techniques to maximize the power efficiency before starting an access. The proposed design is fully scalable and its cost is low. Compared to a conventional IFU design, simulation results show that ABSA saves about 30.3% fetch power consumption, on average. I-Cache employed by ABSA reduces both static and dynamic power consumptions about 85.63% and 66.92%, respectively. Meanwhile the performance degradation is only about 0.97%.