Synthetic-Neuroscore: Using a neuro-AI interface for evaluating generative adversarial networks
Synthetic-Neuroscore: استفاده از یک رابط عصبی هوش مصنوعی برای ارزیابی شبکه های خصمانه تولید -2020
Generative adversarial networks (GANs) are increasingly attracting attention in the computer vision, nat- ural language processing, speech synthesis and similar domains. Arguably the most striking results have been in the area of image synthesis. However, evaluating the performance of GANs is still an open and challenging problem. Existing evaluation metrics primarily measure the dissimilarity between real and generated images using automated statistical methods. They often require large sample sizes for evalua- tion and do not directly reflect human perception of image quality. In this work, we describe an evaluation metric we call Neuroscore , for evaluating the performance of GANs, that more directly reflects psychoperceptual image quality through the utilization of brain signals. Our results show that Neuroscore has superior performance to the current evaluation metrics in that: (1) It is more consistent with human judgment; (2) The evaluation process needs much smaller numbers of samples; and (3) It is able to rank the quality of images on a per GAN basis. A convolutional neural network (CNN) based neuro-AI interface is proposed to predict Neuroscore from GAN-generated images directly without the need for neural responses. Importantly, we show that including neural responses during the training phase of the network can significantly improve the pre- diction capability of the proposed model. Materials related to this work are provided at https://github. com/villawang/Neuro- AI- Interface .
Keywords: Neuroscore | Generative adversarial networks | Neuro-AI interface | Brain-computer interface
Automatic bad channel detection in implantable brain-computer interfaces using multimodal features based on local field potentials and spike signals
تشخیص خودکار کانال بد در رابط های قابل کاشت مغز با کامپیوتر با استفاده از ویژگی های چند حالته بر اساس پتانسیل های محلی و سیگنال های لبه-2020
“Bad channels” in implantable multi-channel recordings bring troubles into the precise quantitative description and analysis of neural signals, especially in the current “big data” era. In this paper, we combine multimodal features based on local field potentials (LFPs) and spike signals to detect bad channels automatically using machine learning. On the basis of 2632 pairs of LFPs and spike recordings acquired from five pigeons, 12 multimodal features are used to quantify each channel’s temporal, frequency, phase and firing-rate properties. We implement seven classifiers in the detection tasks, in which the synthetic minority oversampling technique (SMOTE) system and Fisher weighted Euclidean distance sorting (FWEDS) are used to cope with the class imbalance problem. The results of the two-dimensional scatterplots and classifications demonstrate that correlation coefficient, phase locking value, and coherence have good discriminability. For the multimodal features, almost all the classifiers can obtain high accuracy and bad channel detection rate after the SMOTE operation, in which the Random Forests classifier shows relatively better comprehensive performance (accuracy: 0.9092 � 0.0081, precision: 0.9123 � 0.0100, and recall: 0.9057 � 0.0121). The proposed approach can automatically detect bad channels based on multimodal features, and the results provide valuable references for larger datasets.
Keywords: Bad channel | Multimodal feature | LFP | Spike | Machine learning
سریعترین واسط مغز و کامپیوتر جهان: ترکیب کد EEG2 با یادگیری عمیق
سال انتشار: 2019 - تعداد صفحات فایل pdf انگلیسی: 15 - تعداد صفحات فایل doc فارسی: 29
در این مقاله روش جدیدی بر اساس یادگیری عمیق برای کدگشایی اطلاعات حسیِ حاصل از الکتروانسفالوگرامهایی (EEG) که به صورت غیرتهاجمی ثبت شدهاند، ارائه میدهیم. این روش را میتوان در رابطهای مغز و کامپیوتر (BCI) غیرفعال برای پیشبینی ویژگیهای یک محرک بصری که فرد مشاهده میکند، به کار برد و یا میتوان برای کنترل فعالانهی کاربردهای BCI از آن استفاده کرد. هر دو سناریو مورد آزمایش قرار گرفتند، بدین ترتیب که متوسط نرخ انتقال اطلاعات (ITR) برابر با 701 بیت بر دقیقه برای روش BCI غیرفعال به دست آمد و بهترین سوژه به ITR آنلاین برابر با 1237 بیت بر دقیقه دست یافت. علاوه بر این، امکان تشخیص 500000 محرک بصری مختلف بر اساس تنها 2 ثانیه از اطلاعات EEG با دقت تا 100% را میسر ساخت. هنگامی که این روش در یک BCI خودگام آسنکرون برای هجی کردن به کار برده شد، متوسط نرخ سودمندی برابر با 175 بیت بر دقیقه به دست آمد که متناظر با به طور متوسط 35 حرف بدون خطا در هر دقیقه است. از آنجایی که اطلاعاتی که این روش استخراج میکند، بیش از سه برابرِ سریعترین روش قبلی است، نشان میدهیم که سیگنالهای EEG اطلاعات بیشتری نسبت به مقداری که معمولا فرض میشود، انتقال میدهند. در نهایت یک اثر حداکثر مشاهده کردیم به طوری که محتوای اطلاعات در EEG از آن چیزی که برای کنترل BCI لازم است، فراتر میرود و بنابراین در این مورد بحث میکنیم که آیا تحقیقات BCI به نقطهای رسیدهاند که دیگر نمیتوان عملکرد کنترل BCI بصری غیرتهاجمی را به طور قابل توجهی بهبود بخشید یا خیر.
|مقاله ترجمه شده|
Towards an accessible use of smartphone-based social networks through brain-computer interfaces
به سمت استفاده در دسترس از شبکه های اجتماعی مبتنی بر تلفن های هوشمند از طریق رابط های مغز و کامپیوتر-2019
This study presents an asynchronous P300-based Brain–Computer Interface (BCI) system for controlling social networking features of a smartphone. There are very few BCI studies based on these mobile devices and, to the best of our knowledge, none of them supports networking applications or are focused on an assistive context, failing to test their systems with motor-disabled users. Therefore, the aim of the present study is twofold: (i) to design and develop an asynchronous P300-based BCI system that allows users to control Twitter and Telegram in an Android device; and (ii) to test the usefulness of the developed system with a motor-disabled population in order to meet their daily communication needs. Row-col paradigm (RCP) is used in order to elicitate the P300 potentials in the scalp of the user, which are immediately processed for decoding the user’s intentions. The expert system integrates a decision-making stage that analyzes the attention of the user in real-time, providing a comprehensive and asynchronous control. These intentions are then translated into application commands and sent via Bluetooth to the mobile de- vice, which interprets them and provides visual feedback to the user. During the assessment, both quali- tative and quantitative metrics were obtained, and a comparison among other state-ofthe-art studies was performed as well. The system was tested with 10 healthy control subjects and 18 motor-disabled sub- jects, reaching average online accuracies of 92.3% and 80.6%, respectively. Results suggest that the system allows users to successfully control two socializing features of a smartphone, bridging the accessibility gap in these trending devices. Our proposal could become a useful tool within households, rehabilitation centers or even companies, opening up new ways to support the integration of motor-disabled people, and making an impact in their quality of life by improving personal autonomy and self-dependence.
Keywords: Brain-computer interface (BCI) | Smartphones | Asynchronous control | Social networks | P300 Event-related potentials | Electroencephalography (EEG)
Data mining based approach to study the effect of consumption of caffeinated coffee on the generation of the steady-state visual evoked potential signals
رویکرد مبتنی بر داده کاوی برای بررسی تأثیر مصرف قهوه کافئین دار بر تولید سیگنالهای بالقوه برانگیخته بصری حالت پایدار-2019
The steady-state visual evoked potentials (SSVEP), are elicited at the parieto-occipital region of the cortex when a light source (3.5–75 Hz), flickering at a constant frequency, stimulates the retinal cells. In the last few decades, researchers have reported that caffeine enhances the vigilance and the executive control of visual attention. However, no study has investigated the effect of caffeinated coffee on the SSVEP response, which is used for controlling the brain-computer interface (BCI) devices for rehabilitative applications. The current work proposes a data mining-based approach to gain insight into the alterations in the SSVEP signals after the consumption of caffeinated coffee. Recurrence quantification analysis (RQA) of the electroencephalogram (EEG) signals was employed for this purpose. The EEG signals were acquired at seven frequencies of photic stimuli. The stimuli frequencies were chosen such that they were distributed throughout the EEG frequency bands. The prominent SSVEP signals were identified using the Canonical Correlation Analysis (CCA) method. Several statistical features were extracted from the recurrence plot of the SSVEP signals. Statistical analyses using the t-test and decision tree-based methods helped to select the most relevant features, which were then classified using Automated Neural Network (ANN). The relevant features could be classified with a maximum accuracy of 97%. This supports our hypothesis that the consumption of caffeinated coffee can alter the SSVEP response. In conclusion, utmost care should be taken in selecting the features for designing BCI devices.
Keywords: SSVEP | EEG | Caffeine | Canonical correlation analysis | Recurrence quantification analysis | Multilayer perceptron network
Characterization of phase space trajectories for Brain-Computer Interface
توصیف خط سیر فضایی فاز برای رابط مغز و کامپیوتر-2017
Article history:Received 27 August 2016Received in revised form 4 May 2017 Accepted 15 May 2017Keywords:Brain-Computer Interface (BCI) Electroencephalogram (EEG) Distance series (DS)Moment invariantsPhase space reconstruction (PSR)A new processing framework that allows detailed characterization of the nonlinear dynamics of EEG signals at real-time rates is proposed. In this framework, the phase space trajectory is reconstructed and the underlying dynamics of the brain at different mental states are identiﬁed by analyzing the shape of this trajectory. Two sets of features based on afﬁne-invariant moments and distance series trans- form allow robust estimation of the properties of the phase space trajectory while maintaining real-time performance. We describe the methodological details and practical implementation of the new frame- work and perform experimental veriﬁcation using datasets from BCI competitions II and IV. The results showed excellent performance for using the new features as compared to competition winners and recent research on the same datasets providing best results in Graz2003 dataset and outperforming competition winner in 6 out of 9 subject in Graz2008 dataset. Furthermore, the computation times needed with the new methods were conﬁrmed to permit real-time processing. The combination of more detailed descrip- tion of the nonlinear dynamics of EEG and meeting online processing goals by the new methods offers great potential for several time-critical BCI applications such as prosthetic arm control or mental state monitoring for safety.© 2017 Elsevier Ltd. All rights reserved.1.
Keywords:Brain-Computer Interface (BCI) | Electroencephalogram (EEG) | Distance series (DS) | Moment invariants | Phase space reconstruction (PSR)
The Distribution of Classification Accuracy Over the Whole Head for a Steady State Visual Evoked Potential Based Brain-computer Interface
توزیع به لحاظ دقت طبقه بندی بر کل مجموع ، برای یک حالت پایدار ویژوال پتانسیل برانگیخته بر اساس رابط مغز و کامپیوتر-2017
Brain–computer interfaces (BCIs) system designed using the steady-state visual evoked potential (SSVEP) signal have been widely studied because of their high accuracy of classification and high rates of the information transfer. However, the SSVEP is typically measured over the occipital scalp region (channels O1, O2, and Oz), which makes this type of BCI unsuitable for some patients. We investigated the classification accuracy of SSVEP over the whole scalp, to evaluate the feasibility of building SSVEP-based BCIs that use additional channels. The classification accuracy distribution of the whole scalp increased with the electrode positions closer to the occipital region, and the classification accuracy increased with an increasing number of electroencephalogram data channels.
Keywords: Steady-state visual evoked potential(SSVEP) | Brain–computer interface(BCIs) | Electroencephalogram | Classification accuracy
Human Neuro-Activity for Securing Body Area Networks: Application of Brain-Computer Interfaces to People-Centric Internet of Things
فعالیت های عصبی انسانی برای حفاظت ازمنطقه شبکه های بدن : کاربرد از اینترفیس های مغز-کامپیوتر از افراد مبتنی بر اینترنت اشیاء -2017
A former definition states that a brain-com puter Interface provides a direct communication channel to the brain without the need for mus cles and nerves. With the emergence of wearable and wireless brain-computer interfaces, these systems have evolved to become part of wire less body area networks, offering people-centric applications such as cognitive workload assess ment and detection of selective attention. Cur rently, wireless body area networks are mostly integrated by low-cost devices that, because of their limited hardware resources, cannot gener ate secure random numbers for encryption. This is a critical issue in the context of new Internet of Things device communication and its secu rity. Such devices require securing their com munication, mostly by means of the automatic renewal of the cryptographic keys. In the domain of the people-centric Internet of Things, we pro pose to use wireless brain-computer interfaces as a secure source of entropy, based on neu ro-activity, capable to generate secure keys that outperforms other generation methods. In our approach, current wireless brain-computer inter face technology is an attractive option to offer novel services emerged from novel necessities in the context of the people-centric Internet of Things. Our proposal is an implementation of the human-in-the-loop paradigm, in which devices and humans indistinctly request and offer ser vices to each other for mutual benefit.