Cross-modal interactions between visual understanding and linguistic processing substantially contribute to the remarkable robustness of human language processing. We argue that the formation of cross-modal referential links is a prerequisite for the occurrence of cross-modal interactions between vision and language. In this paper we examine a computational model for a cross-modal reference formation with respect to its robustness against conceptual underspecification in the visual modality. This investigation is motivated by the fact that natural systems are well capable of establishing a cross-modal reference between modalities with different degrees of conceptual specification. In the investigated model, conceptually underspecified context information continues to drive the syntactic disambiguation of verb-centered syntactic ambiguities as long as the visual context contains the situation arity information of the visual scene.

The present study examined whether audiovisual integration of temporal stimulus features in humans can be predicted by the maximum likelihood estimation (MLE) model which is based on the weighting of unisensory cues by their relative reliabilities. In an audiovisual temporal order judgment paradigm, the reliability of the auditory signal was manipulated by Gaussian volume envelopes, introducing varying degrees of temporal uncertainty. While statistically optimal weighting according to the MLE rule was found in half of the participants, the other half consistently overweighted the auditory signal. The results are discussed in terms of a general auditory bias in time perception, interindividual differences, as well as in terms of the conditions and limits of statistically optimal multisensory integration.

A critical parameter of brain-computer interfaces (BCIs) is the number of dimensions a user can control independently. One way to increment this number without increasing the mental effort required to operate the system is to stimulate several sensory modalities simultaneously, and to distinguish brain activity patterns when the user focuses attention to different elements of this multisensory input. In this article we show how shifting attention between simultaneously presented tactile and visual stimuli affects the electrical brain activity of human subjects, and that this signal can be used to augment the control information from the two uni-modal BCI subsystems.

Lesion studies in animals have associated the amygdala mainly with aversive conditioning, both cued and contextual. However, neurophysiological studies indicate that the amygdala has a role in the processing of positive emotions in some kinds of stimulus-reward learning. The aim of the current study was to extend these findings to humans. Functional magnetic resonance imaging was applied to investigate the neural basis of aversive and appetitive conditioning. In the first study, aversive electrical shocks were delivered as either cue or context related. The second study integrated appetitive and aversive conditioning using electric shocks as aversive and monetary rewards as appetitive reinforcers. Differential responses in the amygdala were observed during both cue and contextual conditioning, whereas context conditioning additionally recruited the hippocampus. Amygdala activity was higher for reinforced acoustic stimuli. The findings underline the importance of the amygdala in associative learning, irrespective of the stimulus valence and stimulus type.

Existing methods of obtaining runtime feedback for structure data-layout optimization have several drawbacks, such as large overhead and difficulty composing training sets. As a result, structure data-layout optimization is not widely used. To overcome these drawbacks, a performance monitoring unit (PMU) sampling method was developed with much less overhead and better portability and usability. An algorithm was developed to correct incomplete and inaccurate PMU sampling. With the corrected PMU feedback, a structure data-layout optimizer achieved a 45.1% performance improvement compared to a design without data-layout optimization, which is 97.6% of the performance improvement achieved with instrumented feedback. Calculation of the PMU feedback increased the execution time by 12.3%, compared to the overhead for the instrumented feedback of 341.5%. Tests show that the PMU feedback is efficient and effective for structure data-layout optimization.

The interactions among nano-sized giant magnetoresistances (GMRs) which can affect the biosensor operation, were studied using micromagnetic simulations and micromagnetic analysis. The magnetic field energy, the magnetization distribution and the effective magnetic field were simulated for variable spacer distances, aspect ratios and external magnetic fields to study the interactions among the GMR sensors. And the spacer distance should be at least 1000 nm to eliminate interference between 1000 nm× 100 nm×9 nm or larger GMRs in an array, while the aspect ratio should be more than 10 to improve the sensor sensitivity and stability. Micromagnetic energy theory is used to explain the simulation results.

An ultra-low power complementary metal-oxide-semiconductor (CMOS) front-end readout ASIC was developed for a portable digital radiation detector. The ASIC having a charge sensitive amplifier and a semi-Gaussian pulse-shaper was produced using the CSMC 0.5 μm DPDM process. The ENC noise of 363 e at 0 pF with a noise slope of 23 e/pF complies with the stringent low noise requirements. The peaking time was 250 ns at a 100 mV/fC conversion gain (detector capacitance is 20 pF). By operating this front-end readout ASIC in the weak inversion region, the ultra-low power dissipation is only 0.1 mW/channel (3.0 V). Simulations and test results suggest that this design gives lower power consumption than the front-end readout ASICs working in the strong inversion and is appropriate for the portable digital radiation detectors.

The power consumption by the data cache is important in DSP designs. This study presents an enhanced branch access LRU-SEQ (EBA-LRU-SEQ) policy for data caches in DSP designs to reduce the power consumption. The design is based on the LRU policy with embedded prefetch table to provide branch access. Tests show that the EBA-LRU-SEQ policy reduces the data cache power consumption to 54% of a system with no power control.

A method was developed to estimate EEPROM device life based on the consistency for breakdown charge, Q_BD, for constant voltage time dependent dielectric breakdown (TDDB) and constant current TDDB stress tests. Although an EEPROM works with a constant voltage, Q_BD for the tunnel oxide can be extracted using a constant current TDDB. Once the charge through the tunnel oxide, ΔQ_FG, is measured, the lower limit of the EEPROM life can be related to Q_BD/ΔQ_FG. The method is reached by erase/write cycle tests on an EEPROM.

Modeling of Schottky diodes in the CMOS process is a key step in ultra-high frequency (UHF) radio frequency identification (RFID) transponder designs. Accurate Schottky diode models need both DC and RF models. Conventional DC models of the Schottky diode fail to predict the forward leakage current, which is crucial for precise simulation results. This paper presents a Schottky diode model with an additional diode which gives the correct forward leakage current. The RF model of the Schottky diode is constructed based on the measured S-parameters. Then, an on-chip de-embedding process is needed to remove the parasitics due to the pads and interconnection lines in the S-parameter test. A flexible “open-through” on-chip de-embedding method is proposed which only requires an “open” dummy and a “through” dummy, with all the lumped and distributed parasitics equivalent to two-port networks to give sufficient high-frequency de-embedding accuracy. By the help of this de-embedding method and the new DC model, the accuracy of the established diode model could be guaranteed. The Schottky diode model is verified by comparison between measurements and simulations and successfully applied to an RFID transponder design.

Engineering-oriented simulations of quantum mechanical tunneling are often based on density-gradient (DG) theory. This paper presents an analytical solution to the DG equation for quantum tunneling through an ultra-thin oxide in a MOS capacitor with an n+ poly-silicon gate obtained using the method of matched asymptotic expansions. Tunneling boundary conditions extend the approximation into the entire region of the poly-silicon gate, oxide barrier, and substrate. An analytical solution in the form of an asymptotic series is obtained in each region by treating each part of the domain as a separate singular perturbation problem. The solutions are then combined through ‘matching’ to obtain an approximate solution for the whole domain. Analytical formulae are given for the electrostatic potential and the electron density profiles. The results capture the features of the quantum effects which are quite different from classical physics predictions. The analytical results compare well with exact numerical solutions over a broad range of voltages and different oxide thicknesses. The analytical results predict the enhancement of the quantum tunneling effect as the oxide thickness is reduced.

A stronger canonical model was developed to improve the performance of automatic pronunciation evaluations. Three different strategies were investigated with speaker adaptive training to normalize variations among speakers, minimum phone error training to identify easily confused phones and maximum likelihood linear regression (MLLR) adaptation to compensate for accent variations between native and non-native speakers. The three schemes were combined to improve the correlation coefficient between machine scores and human scores from 0.651 to 0.679 on the sentence level and from 0.788 to 0.822 on the speaker level.

The idea of using a one-time-one-key design has been widely applied in conventional cryptography. With the security theory of conventional cryptology, encryption algorithms are made public while all the secrets are encoded only in the keys. This paper applies chaos theory to conventional cryptography to develop a one-time-one-algorithm design. A general theory is given to generate the clock key, substitution box, permutation box and operational sign functions for a one-time-one-algorithm scheme. This scheme is then implemented in a system to manage the tradeoff between speed and the security of the encryption algorithm.

The cluster global time can change in wireless sensor networks due to leader re-elections and node disabilities between two successive synchronizations, which will affect temporal relationships. This paper analyzes cluster global time continuity, using global time change models for the node dynamics. The results prove that defining the global time using the cluster average time (AGT) is more stable than defining it using a single node's local time (SGT). With normally distributed clock-parameter assumptions, the AGT change bounds are at most 70.7% of those for the SGT's. The impacts of the initial phase and frequency skew distributions on the global time continuity are also investigated to show that the initial phase variations may strongly influence the continuity. Simulations show that the AGT is more stable with less disabled nodes or larger clusters. The appropriate cluster size is 20–40 when there are less than 15 disabled nodes.

The histogram of oriented gradient has been successfully applied in many research fields with excellent performance especially in pedestrian detection. However, the method has rarely been applied to face recognition. Aimed to develop a fast and efficient new feature for face recognition, the original HOG and its variations were applied to evaluate the effects of different factors. An information theory-based criterion was also developed to evaluate the potential classification power of different features. Comparative experiments show that even with a relatively simple feature descriptor, the proposed HOG feature achieves almost the same recognition rate with much lower computational time than the widely used Gabor feature on the FRGC and CAS-PEAL databases.