Novel Online Censoring Based Learning Algorithm For Complex-Valued Big Data Streams

Güvenç, Buket Çolak; Eren, Yusuf; Mengüç, Engin Cemal

doi:10.1109/siu55565.2022.9864761

Cited by 2 publications

(1 citation statement)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To overcome mentioned issues, the online censoring (OC) strategy has been proposed in [17] and has been a significant focus point in several applications such as the processing of big data streams [17], [18], [19], the modeling of pathological hand tremor signals [20], [21], the prediction of wind speed [22], the design of complex-valued adaptive filters [23] and the cancellation of the AF [4], etc. The OC strategy integrated into an adaptive algorithm provides to censor the non-innovative data (so-called non-informative or redundant data) in an online manner by taking into account a statistical censoring rule [17].…”

Section: Introductionmentioning

confidence: 99%

An acoustic feedback canceler based on probe noise and informative data for hearing aids

Eren,

Güvenç,

Mengüç

2023

SIViP

Self Cite

View full text Add to dashboard Cite

One of the important issues encountered in acoustic feedback canceller (AFC) systems is the bias occurring in the estimation of the acoustic feedback (AF) path. In the literature, a common solution is to inject the probe noise into the loudspeaker to reduce the correlation between the input signal and the loudspeaker signal, that is, to remove the bias occurring in the estimation of the AF path. This approach significantly improves the performance of the AFC systems. However, when considering high filter orders and large-scale input signals, such systems designed by probe noise injection suffer from the data processing cost, and thus their real-time implementations on a chip generally become challenging.To this end, in this paper, we propose an AFC system using jointly probe noise and informative data, named the probe noise-online censoring-least mean square (PN-OC-LMS) based AFC. Thanks to both the OC mechanism and the probe noise injection approach, the proposed PN-OC-LMS based AFC system significantly not only reduces the data processing cost but also effectively removes the bias effect in the AF path. The mentioned effectiveness of the proposed AFC system is supported by two comprehensive different experiments on a real-world AF path measured from a behind-the-ear hearing aid. As a result, in the proposed AFC system, the use of both informative data and the injection of probe noise will enable the design of a long-term working wearable hearing aid device that provides low data processing cost and correctly estimates the AF path.

show abstract

Section: Introductionmentioning

confidence: 99%

An acoustic feedback canceler based on probe noise and informative data for hearing aids

Eren,

Güvenç,

Mengüç

2023

SIViP

Self Cite

View full text Add to dashboard Cite

show abstract

An acoustic feedback canceller using probe noise and informative data

Eren,

Güvenç,

Mengüç

2023

Preprint

View full text Add to dashboard Cite

One of the important issues encountered in acoustic feedback canceller (AFC) systems is the bias occurring in the estimation of the acoustic feedback (AF) path. In the literature, a common solution is to inject the probe noise into the loudspeaker to reduce the correlation between the input signal and the loudspeaker signal, that is, to remove the bias occurring in the estimation of the AF path. This approach significantly improves the performance of the AFC systems. However, when considering high filter orders and large-scale input signals, such systems designed by probe noise injection suffer from the data processing cost, and thus their real-time implementations on a chip generally become challenging. To this end, in this paper, we propose an AFC system using jointly probe noise and informative data, named the probe noise-online censoring-least mean square (PN-OC-LMS) based AFC. Thanks to both the OC mechanism and the probe noise injection approach, the proposed PN-OC-LMS based AFC system significantly not only reduces the data processing cost but also effectively removes the bias effect in the AF path. The mentioned effectiveness of the proposed AFC system is supported by two comprehensive different experiments on a real-world AF path measured from a behind-the-ear hearing aid. As a result, in the proposed AFC system, the use of both informative data and the injection of probe noise will enable the design of a long-term working wearable hearing aid device that provide slow data processing cost and correctly estimates the AF path.

show abstract

Novel Online Censoring Based Learning Algorithm For Complex-Valued Big Data Streams

Cited by 2 publications

References 0 publications

An acoustic feedback canceler based on probe noise and informative data for hearing aids

An acoustic feedback canceler based on probe noise and informative data for hearing aids

An acoustic feedback canceller using probe noise and informative data

Contact Info

Product

Resources

About