Spoken accents severely degrade the performance of automatic speech recognition (ASR) systems. Domain adversarial training (DAT) is widely adopted for generating domain-invariant features to reduce the influence of ac...
详细信息
ISBN:
(纸本)9798350397970
Spoken accents severely degrade the performance of automatic speech recognition (ASR) systems. Domain adversarial training (DAT) is widely adopted for generating domain-invariant features to reduce the influence of accents. However, the generated features trained by DAT are still maintaining some accent discrimination information, limiting the ASR performance. In addition, the features generated by DAT of each accent have different degrees of residual accent discriminant information. In this paper, we propose an adaptive attention network with DAT to further eliminate the influence of retaining accent information in features generated by DAT. We employ the adaptive attention module to transform the encoder output to a more general representation. Experiments on the AESRC2020 dataset show that the proposed method can achieve satisfactory performance improvements on seen and unseen accent when the correct accent information is still preserved in the output of the encoder.
The novel coronavirus (nCoV-19) was first detected in December 2019. It had spread worldwide and was declared coronavirus disease (COVID-19) pandemic by March 2020. Patients presented with a wide range of symptoms aff...
详细信息
Dynamic programming is a fundamental algorithm that can be found in our daily lives easily. One of the dynamic programming algorithm implementations consists of solving the 0/1 knapsack problem. A 0/1 knapsack problem...
Dynamic programming is a fundamental algorithm that can be found in our daily lives easily. One of the dynamic programming algorithm implementations consists of solving the 0/1 knapsack problem. A 0/1 knapsack problem can be seen from industrial production cost. It is prevalent that a production cost has to be as efficient as possible, but the expectation is to get the proceeds of the products higher. Thus, the dynamic programming algorithm can be implemented to solve the diverse knapsack problem, one of which is the 0/1 knapsack problem, which would be the main focus of this paper. The implementation was implemented using C language. This paper was created as an early implementation algorithm using a Dynamic program algorithm applied to an Automatic Identification System (AIS) dataset.
Dual-encoder structure successfully utilizes two language-specific encoders (LSEs) for code-switching speech recognition. Because LSEs are initialized by two pre-trained language-specific models (LSMs), the dual-encod...
详细信息
Students’ interactions while solving problems in learning environments (i.e. log data) are often used to support students’ learning. For example, researchers use log data to develop systems that can provide students...
详细信息
As the availability of data is increasing everyday, the need to reflect on how to make these data meaningful and impactful becomes vital. Current data paradigms have provided data life cycles that often focus on data ...
详细信息
This work studied message communications on patient portals and examined both the longitudinal trends and the correlations with characteristics of message senders. We analyzed over 5.6 million secure messages sent on ...
This work studied message communications on patient portals and examined both the longitudinal trends and the correlations with characteristics of message senders. We analyzed over 5.6 million secure messages sent on the Mayo Clinic patient portal between February 18, 2010, and December 31, 2017. We studied the longitudinal changes in the number of portal messages, patient senders’ demographics and medical conditions (PheCodes), and provider senders’ care settings (e.g., primary or specialty) and practice roles (e.g., physician, nurse practitioner, and registered nurses). When compared to non-message-senders, patient message senders had a significantly higher proportion of the demographics: age 41-60, female, married, white, and English-speaking. From 2010-2017, an individual patient sent an average of 9.8 messages per person while a provider sent 418.4. The average number of PheCodes for all patients regardless of portal usage increased from 7.5 +/-6.9 in 2010 to 10.7 +/- 10.1 in 2017. The Pearson correlation coefficient between average PheCodes per patient and average messages per patient was 0.273 (p < 0.0001). Physicians were the largest proportion of message composers in both primary and specialty care (36.20% of primary, 37.54% of specialty). Starting 2013 onwards, specialty providers comprised the majority of portal providers while primary care providers remained stable around 20-22%. Our results show that patient portals are playing an increasingly significant role in supporting patient-provider communications. The longitudinal growth also sheds light on the possible challenge of communication overload for providers and the healthcare system.
One major way that people engage in adaptive problem solving is by imitating others’ solutions. Prominent simulation models have found imperfect imitation advantageous, but the interactions between copying amount and...
详细信息
One major way that people engage in adaptive problem solving is by imitating others’ solutions. Prominent simulation models have found imperfect imitation advantageous, but the interactions between copying amount and other prevalent aspects of social learning strategies have been underexplored. Here, we explore the consequences for a group when its members engage in strategies with different degrees of copying, solving search problems of varying complexity, in different network topologies that affect the solutions visible to each member. Using a computational model of collective problem solving, we demonstrate that the advantage of partial copying is robust across these conditions, arising from its ability to maintain diversity. Partial copying delays convergence generally but especially in globally connected networks, which are typically associated with diversity loss, allowing more exploration of a problem space. We show that a moderate amount of diversity maintenance is optimal and strategies can be adjusted to find that sweet spot.
Large language models (LLMs) are increasingly utilized in healthcare applications. However, their deployment in clinical practice raises significant safety concerns, including the potential spread of harmful informati...
详细信息
Study on the identification and classification of fish is challenging and valuable because of its role in advancing the marine and agricultural fields. This research has benefits interms of monitoring fish populations...
详细信息
ISBN:
(纸本)9781665473286
Study on the identification and classification of fish is challenging and valuable because of its role in advancing the marine and agricultural fields. This research has benefits interms of monitoring fish populations and ecosystems in a particular area. Furthermore, this research helps monitor fish that are considered threatened or endangered so that it makes iteasier to map prohibited areas for fishing. This research aims to know performance of MobileNetV2 and VGG16 with parameter tuning process by identifying the value of batch size, epoch, learning rate, and optimizer for fish image dataset. The proposed research phase consists of five main stages, including experimental setup, dataset construction, dataset preprocessing, dataset training and modelling and evaluation. As the result, VGG16 obtained the highest accuracy value. For VGG16 without fine-tuning, the testing accuracy is 98.07%. For VGG16 with fine-tuning, the testing accuracy is 96.56%.
暂无评论