管理

流量統計

  • 52270總讀取次數:
  • 2讀取次數(今天):
  • 16讀取次數(昨天):
  • 98讀取次數(上週):
  • 23018總參觀人數:
  • 1今天的參觀人數:
  • 0目前在線人數:

訂閱中華機率統計學會RSS

統計所演講訊息(rss)

The deep learning technology in cochlear implant

Mon, 25 Dec 2017 14:26:31 +0800
 

Abstract

    Cochlear implants (CIs) are surgically implanted electronic devices that provide a sense of sound in patients with profound to severe hearing loss. The considerable progress of CI technologies in the past three decades has enabled many CI users to enjoy a high level of speech understanding in quiet. For most CI users, however, understanding speech in noisy environments remains a challenge. In this talk, I will present the deep learning based noise reduction (NR) approach, which has been demonstrated its effectiveness for improving the speech intelligibility for CI recipients. Experimental results indicated that deep learning based NR yields higher intelligibility scores than conventional approaches for Mandarin-speaking listeners, suggesting that DDAE NR could potentially be integrated into a CI processor to overcome speech perception degradation caused by noise.

>
Read more ...

Optimal Group Testing Design with Cost Considerations and Dilution Effects

Tue, 16 Jan 2018 16:07:45 +0800
 

Abstract

    A group testing study involves collecting samples from multiple individuals, pooling them, and testing them as a group. A realistic cost model for such a study should consider the costs both for collecting the samples, and for testing the assays. One main goal of group testing is to estimate the prevalence of a disease, which can be biased due to misspecified nominal values of test sensitivity and specificity. An efficient design should accommodate such inaccuracies. In our series of works, we derive locally optimal designs in this setting, and characterize their theoretical properties. We also provide a guaranteed algorithm for constructing the designs on discrete design spaces. Several simulated examples based on a chlamydia study in the USA show that the proposed designs have high efficiency, and are not strongly sensitive to the working parameter specification that is used to obtain the locally optimal design. (Work done jointly with M.-N. L. Huang, K. Shedden, and W. K. Wong.)

 

Key words and phrases: bias-variance trade-off; budget constrained design; dilution effect model; D-optimality; group testing>
Read more ...

Towards Brain Decoding for Real-World Alertness Estimation

Tue, 16 Jan 2018 17:22:43 +0800
 

Abstract

    A brain-computer interface (BCI) allows human to communicate with a computer by thoughts. Recent advances in brain decoding have shown the capability of BCIs in monitoring physiological and cognitive state of the brain, including alertness. Since drowsy driving has been an urgent issue in vehicle safety that causes numerous deaths and injuries, BCIs based on non-invasive electroencephalogram (EEG) are developed to monitor drivers’ alertness continuously and instantaneously. Nonetheless, on the pathway of transitioning laboratory-oriented BCI into real-world applications, there are major challenges that limit the usability and convenience for alertness estimation (AE). To completely understand the association between human EEG and alertness, this study employed a large-scale dataset collected from simulated driving experiments with a lane-keeping task and EEG recordings. An AE-BCI that acquires EEG from only non-hair-bearing (NHB) areas was proposed to maximize the comfort and convenience. The performance of the NHB AE-BCI was validated and compared with that using whole-scalp EEG, showing no significant difference in the accuracy of alert/non-alert classification. In addition, a subject-transfer framework that leverages large-scale existing data from other subjects was proposed to reduce the calibration time of an AE-BCI. Alert baseline data were involved to enhance the efficiency of subject-to-subject model transfer. The subject-transfer approach significantly reduced the calibration time of the AE-BCI, exhibiting the potential in facilitating plug-and-play brain decoding for real-world BCI applications. Overall, this study presents the contributions to developing an AE-BCI for real-world use with maximal usability and convenience. The methodologies and findings could further catalyze the exploration of real-world BCIs in more applications.

>
Read more ...