In this paper, we present a novel Deep Neural Network-based indoor localization method that estimates the position of a Bluetooth Low Energy (BLE) transmitter (tag) by using the received signals’ characteristics at multiple Anchor Points (APs). We use the received signal strength indicator (RSSI) value and the in-phase and quadrature-phase (IQ) components of the received BLE signals at a single time instance to simultaneously estimate the angle of arrival (AoA) at all APs. Through supervised learning on simulated data, various machine learning (ML) architectures are trained to perform AoA estimation using varying subsets of anchor points. In the final stage of the system, the estimated AoA values are fed to a positioning engine which uses the least squares (LS) algorithm to estimate the position of the tag. The proposed architectures are trained and rigorously tested on several simulated room scenarios and are shown to achieve a localization accuracy of 70 cm. Moreover, the proposed systems possess generalization capabilities by being robust to modifications in the room’s content or anchors’ configuration. Additionally, some of the proposed architectures have the ability to distribute the computational load over the APs.
The fingerprinting technique is a popular approach to reveal location of persons, instruments or devices in an indoor environment. Typically based on signal strength measurement, a power level map is created first in the learning phase to align with measured values in the inference. Second, the location is determined by taking the point for which the recorded received power level is closest to the power level actually measured. The biggest limit of this technique is the reliability of power measurements, which may lack accuracy in many wireless systems. To this end, this work extends the power level measurement by using multiple anchors and multiple radio channels and, consequently, considers different approaches to aligning the actual measurements with the recorded values. The dataset is available online. This article focuses on the very popular radio technology Bluetooth Low Energy to explore the possible improvement of the system accuracy through different machine learning approaches. It shows how the accuracy–complexity trade-off influences the possible candidate algorithms on an example of three-channel Bluetooth received signal strength based fingerprinting in a one dimensional environment with four static anchors and in a two dimensional environment with the same set of anchors. We provide a literature survey to identify the machine learning algorithms applied in the literature to show that the studies available can not be compared directly. Then, we implement and analyze the performance of four most popular supervised learning techniques, namely k Nearest Neighbors, Support Vector Machines, Random Forest, and Artificial Neural Network. In our scenario, the most promising machine learning technique being the Random Forest with classification accuracy over 99%.
16 March 2023 in IEEE Transactions on Intelligent Transportation Systems
Forecasting vessel locations is of major importance in the maritime domain, with applications in safety, logistics, etc. Nowadays, vessel tracking has become possible largely due to the increased GPS-based data availability. This paper introduces a novel Vessel Location Forecasting (VLF) framework, based on Long-Short Term Memory (LSTM) Neural Networks, aiming to perform effective location forecasting in time horizons up to 60 minutes, even for vessels not recorded in the past. The proposed VLF framework is specially designed for handling vessel data by addressing some major GPS-related obstacles including variable sampling rate, sparse trajectories, and noise contained in such data. Our framework also learns by incorporating a novel trajectory data augmentation method to improve its predictive power. We validate VLF framework using three real-word datasets of vessels moving in different sea areas, comparing with various methods, and examining several aspects. Results prove VLF framework’s generic nature, robustness regarding parameter changes, and superiority against state of the art in terms of prediction accuracy (higher than 30%) and computational effort.
Automatically synthesizing dance motion sequences is an increasingly popular research task in the broader field of human motion analysis. Recent approaches have mostly used recurrent neural networks (RNNs), which are known to suffer from prediction error accumulation, usually limiting models to synthesize short choreographies of less than 100 poses. In this paper we present a multimodal convolutional autoencoder that combines 2D skeletal and audio information by employing an attention-based feature fusion mechanism, capable of generating novel dance motion sequences of arbitrary length. We first validate the ability of our system to capture the temporal context of dancing in a unimodal setting, by considering only skeletal features as input. According to 1440 rating answers provided by 24 participants in our initial user-study, we show that the optimal performance was presented by the model that was trained with input sequences of 500 poses. Based on this outcome, we train the proposed multimodal architecture with two different approaches, namely teacher-forcing and self-supervised curriculum learning, to deal with the autoregressive error accumulation phenomenon. In our evaluation campaign, we generate 1800 sequences and compare our method against two state-of-the-art approaches. Through qualitative and quantitative experiments we demonstrate the improvements introduced by the proposed multimodal architecture in terms of realism, motion diversity and multimodality, reducing the Fréchet Inception Distance (FID) metric value by 0.39. Subjective results confirm the effectiveness of our approach to synthesize diverse dance motion sequences, reporting a 6% increase in style consistency preference according to 1800 answers provided by 45 evaluators.
The development of methods that combine different sources of information for the purposes of medical diagnosis is an essential challenge in the field of medical informatics. In this context, we introduce a machine-learning framework for automatic voice pathology classification and, in particular, a modular deep learning architecture that classifies voice signals stemming from four types of voice disorders. To this end, we design a multimodal deep learning architecture that fuses medical metadata with voice signals. Our classifier is a combination of fully convolutional and feed-forward sub-networks that simultaneously process low-level and mid-level features which are extracted from acoustic signals of varying duration and medical records, respectively. A key objective of our study is to develop an architecture that is capable of processing voice samples of varying duration, in order to enhance the system’s learning and inference capabilities. Our research also focuses on overcoming performance limitations of neural networks that stem from the lack of extensive volumes of training data. We therefore investigate problem specific augmentation techniques based on the feature sequence segmentation and coloured noise injection and we show that the proposed method gives state-of-the-art results, achieving 64.4% classification accuracy, compared to the 63.5% classification score of the best performing method of the 2019 FEMH data challenge.
8 August 2014 in International Journal of Information Security