Diabetes is considered one of the life-threatening diseases in the world, which needs regular monitoring of blood glucose levels. In this article, we developed a portable system that makes near-infrared spectroscopy (NIRS) technology available to non-professionals through a mobile application and a specially-made enclosure. It overcomes the shortcomings of traditional spectroscopy systems, such as large volume, high cost, complicated operation, and difficulty in online detection. To verify the feasibility of NIRS in noninvasive blood glucose concentration detection, after the pretreatment of the acquired original spectra, we compared two different feature extraction algorithms of synergy interval (Si) and genetic algorithm (GA). On this basis, two quantitative prediction models of partial least squares (PLS) and extreme learning machine (ELM) were established. The experimental results showed the model based on the combination of Si and GA and ELM (i.e., Si-GA-ELM model) as the most accurate among the selected models. At the same time, the prediction accuracy of the spectral waveband was higher than that of the full. To further overcome the difficulty of establishing a finite sample data model and reduce the influence of individual differences, the model transfer method TrAdaBoost was used to enhance the accuracy and stability of our model. The final experimental results show that the NIR spectrometer used is portable and light and can be encased as a handheld device form. Computation models combining machine learning and chemometric methods make the estimated blood glucose more feasible, which is an innovative work in noninvasive blood glucose measurement fields.

An Accurate Noninvasive Blood Glucose Measurement System Using Portable Near-Infrared Spectrometer and Transfer Learning Framework

Biological brains have a natural capacity for resolving certain classification tasks. Studies on biologically plausible spiking neurons, architectures and mechanisms of artificial neural systems that closely match biological observations while giving high classification performance are gaining momentum. Spiking neural P systems (SN P systems) are a class of membrane computing models and third-generation neural networks that are based on the behavior of biological neural cells and have been used in various engineering applications. Furthermore, SN P systems are characterized by a highly flexible structure that enables the design of a machine learning algorithm by mimicking the structure and behavior of biological cells without the over-simplification present in neural networks. Based on this aspect, this paper proposes a novel type of SN P system, namely, layered SN P system (LSN P system), to solve classification problems by supervised learning. The proposed LSN P system consists of a multi-layer network containing multiple weighted fuzzy SN P systems with adaptive weight adjustment rules. The proposed system employs specific ascending dimension techniques and a selection method of output neurons for classification problems. The experimental results obtained using benchmark datasets from the UCI machine learning repository and MNIST dataset demonstrated the feasibility and effectiveness of the proposed LSN P system. More importantly, the proposed LSN P system presents the first SN P system that demonstrates sufficient performance for use in addressing real-world classification problems. 

A Layered Spiking Neural System for Classification Problems

Deep neural networks have become increasingly popular in radar micro-Doppler classification; yet, a key challenge, which has limited potential gains, is the lack of large amounts of measured data that can facilitate the design of deeper networks with greater robustness and performance. Several approaches have been proposed in the literature to address this problem, such as unsupervised pre-training and transfer learning from optical imagery or synthetic RF data. This work investigates an alternative approach to training which involves exploitation of “datasets of opportunity" micro-Doppler datasets collected using other RF sensors, which may be of a different frequency, bandwidth or waveform - for the purposes of training. Specifically, this work compares in detail the cross-frequency training degradation incurred for several different training approaches and deep neural network (DNN) architectures. Results show a 70% drop in classification accuracy when the RF sensors for pre-training, fine-tuning, and testing are different, and a 15% degradation when only the pre-training data is different, but the fine-tuning and test data are from the same sensor. By using generative adversarial networks (GANs), a large amount of synthetic data is generated for pre-training. Results show that cross-frequency performance degradation is reduced by 50% when kinematically-sifted GAN-synthesized signatures are used in pre-training.

/pdf/cross-frequency-training-with-adversarial-learning-for-radar-lt7mz3v4gb.pdf

Cross-frequency training with adversarial learning for radar micro-Doppler signature classification (Rising Researcher)

The usage of a commercial 24-GHz frequency-modulated continuous-wave radar sensor to retrieve sound signals is investigated in this letter. Thanks to the great phase measurement capabilities of current radar sensors, low-frequency audio signals can be recovered by measuring the vibrations they induce in objects due to their pressure wave nature. To prove the concept, a sound emitted by a speaker is recorded by analyzing the small-scale displacement of a reflective surface using a radar sensor originally devised for automotive and unmanned aerial vehicle (UAV) applications at the 24-GHz industrial, scientific, and medical band.

Microwave Microphone Using a General Purpose 24-GHz FMCW Radar

At present, specific voice control has gradually become an important means for 5G-Internet-of-Things-aided industrial control systems, such as controlling the operation and adjustment of industrial Internet of Things equipment through telephone voice of the controller. However, the security of specific voice control system needs to be improved, because the voice cloning technology based on transfer learning can easily simulate the voice of the controller, which may lead to industrial accidents and other potential security risks. Therefore, this article mainly aims to study and understand the principle of voice cloning attack technology, putting forward a voice clone attack method, in order to prepare for the construction of a specific voice recognition system in the future. At present, the key technology of voice cloning attack is how to solve the problem that the target speaker's personalized speech with high quality cannot be synthesized under small samples. In fact, voice cloning is a very challenging problem because speech is more difficult to be represented in the hidden space of the model. We propose a transductive voice transfer learning method to learn the predictive function from the source domain and fine-tune in the target domain adaptively. The target learning task and the source learning task are both synthesizing speech signals from the given audio, while the datasets of both domains are different. By adding different penalty values to each instances and minimizing the expected risk, an optimal precise model can be learned. In addition, an evaluation method to verify the audio similarity of the target speaker was given to show the similarity between the synthesized audio and the original audio. Many details of the experimental results show that our method can effectively synthesize the speech of the target speaker with small samples.

Voice-Transfer Attacking on Industrial Voice Control Systems in 5G-Aided IIoT Domain

We investigated the feasibility of using Doppler radar to recognize human voices by capturing the micro-Doppler signatures of vibrations from the larynx and mouth. The signatures produced through the vibrations of a human being’s vocal cords generate unique micro-Doppler signatures, depending on the letters pronounced. These can then be used to classify and recognize different words and letters. In this paper, we could successfully capture echo signals using the Doppler radar when a human subject spoke seven musical notes from Do to Ti and alphabet letters from $\boldsymbol {A}$ to $\boldsymbol {Z}$ . Spectrogram analysis was conducted for classification purposes, and the deep convolutional neural networks employed could classify the 26 letters to an accuracy of 94%. To overcome the deficiency of the measured data and improve the classification accuracy, transfer learning was introduced. Using the VGG-16 model, its accuracy was improved up to 97%. Additional experiments were conducted to ascertain the radar’s capability to detect the human voice through a barrier between the human and the radar. In this paper, we demonstrated the possibility of remote voice recognition using Doppler information, with or without a barrier.

Through-Wall Remote Human Voice Recognition Using Doppler Radar With Transfer Learning

We introduce Alexa Arena, a user-centric simulation platform for Embodied AI (EAI) research. Alexa Arena provides a variety of multi-room layouts and interactable objects, for the creation of human-robot interaction (HRI) missions. With user-friendly graphics and control mechanisms, Alexa Arena supports the development of gamified robotic tasks readily accessible to general human users, thus opening a new venue for high-efficiency HRI data collection and EAI system evaluation. Along with the platform, we introduce a dialog-enabled instruction-following benchmark and provide baseline results for it. We make Alexa Arena publicly available to facilitate research in building generalizable and assistive embodied agents.

Rohan Khanna

Papers

Through-Wall Remote Human Voice Recognition Using Doppler Radar With Transfer Learning

Alexa Arena: A User-Centric Interactive Platform for Embodied AI