The properties of the 2,570 Chinese characters explicitly taught in Chinese elementary schools were systematically investigated, including types of characters, visual complexity, spatial structure, phonetic regularity and consistency, semantic transparency, independent and bound components, and phonetic and semantic families. Among the findings are that the visual complexity, phonetic regularity, and semantic transparency of the Chinese characters taught in elementary school increase from the early grades to the later grades: Characters introduced in the 1st or 2nd grade typically contain fewer strokes, but are less likely to be regular or transparent, than characters introduced in the 5th or 6th grade. The inverse relation holds when characters are stratified by frequency. Low-frequency characters tend to be visually complex, phonetically regular, and semantically transparent whereas high-frequency characters tend to be the opposite. Combined with other findings, the analysis suggests that written Chinese has a logic that children can understand and use.

This study investigated the development of phonetic awareness, meaning insight into the structure and function of the component of Chinese characters that gives a clue to pronunciation. Participants were 113 Chinese 2nd, 4th, and 6th graders enrolled in a working-class Beijing, China elementary school. The children's task was to represent the pronunciation of 60 semantic phonetic compound characters. As anticipated, both character familiarity and character regularity strongly influenced performance. Children as young as 2nd graders are better able to represent the pronunciation of regular characters than irregular characters or characters with bound phonetics. Phonetic awareness continues to develop over the elementary school years, as is shown by the increasing influence of phonetic regularity on the performance of children in higher grades and the increasing percentage of phonetic-related errors among older children.

This study investigated whether children can use partial information to learn the pronunciations of Chinese characters. Participants were 49 2nd graders and 56 4th graders whose home language was Mandarin and 75 2nd graders and 93 4th graders whose home language was Cantonese. Children had 2 trials to learn the Mandarin pronunciations of 28 unfamiliar compound characters of 4 types. Children learned to pronounce more regular characters, which contain full information about pronunciation, and more tone-different and onset-different characters, which contain partial information about pronunciation, than characters with unknown phonetic components, which contain no information about pronunciation. Mandarin-speaking children learned more pronunciations than Cantonese-speaking children.

In this paper, we apply machine learning to distributed private data owned by multiple data owners, entities with access to non-overlapping training datasets. We use noisy, differentially-private gradients to minimize the fitness cost of the machine learning model using stochastic gradient descent. We quantify the quality of the trained model, using the fitness cost, as a function of privacy budget and size of the distributed datasets to capture the trade-off between privacy and utility in machine learning. This way, we can predict the outcome of collaboration among privacy-aware data owners prior to executing potentially computationally-expensive machine learning algorithms. Particularly, we show that the difference between the fitness of the trained machine learning model using differentially-private gradient queries and the fitness of the trained machine model in the absence of any privacy concerns is inversely proportional to the size of the training datasets squared and the privacy budget squared. We successfully validate the performance prediction with the actual performance of the proposed privacy-aware learning algorithms, applied to: financial datasets for determining interest rates of loans using regression; and detecting credit card frauds using support vector machines.

This paper describes the development of a phasesensitive laser probe with fast mechanical scan for RF surface and bulk acoustic wave (SAW/BAW) devices. The Sagnac interferometer composed of micro-optic elements was introduced for the selective detection of RF vertical motion associated with RF SAW/BAW propagation and vibration. A high-pass characteristic of the interferometer makes the measurement very insensitive to low-frequency vibration. This feature allows us to apply the fast mechanical scan to the interferometric measurement without badly sacrificing its SNR and spatial resolution. The system was applied to the visualization of a field pattern on the vibrating surface of an RF BAW resonator operating in the 2 GHz range. The field pattern was obtained in 17 min as a 2-D image (500 × 750 pixel with 0.4 μm resolution and SNR of 40 dB). The system was also applied to the characterization of an RF SAW resonator operating in the 1 GHz range, and the applicability of the system was demonstrated.