Empirical comparison of deep learning methods for EEG decoding.

Many human activities consist of physical gestures that tend to be performed in certain sequences. Wearable inertial sensor data have as a consequence been employed to automatically detect human activities, lately predominantly with deep learning methods. This article focuses on the necessity of recurrent layers—more specifically Long Short-Term Memory (LSTM) layers—in common Deep Learning architectures for Human Activity Recognition (HAR). Our experimental pipeline investigates the effects of employing none, one, or two LSTM layers, as well as different layers' sizes, within the popular DeepConvLSTM architecture. We evaluate the architecture's performance on five well-known activity recognition datasets and provide an in-depth analysis of the per-class results, showing trends which type of activities or datasets profit the most from the removal of LSTM layers. For 4 out of 5 datasets, an altered architecture with one LSTM layer produces the best prediction results. In our previous work we already investigated the impact of a 2-layered LSTM when dealing with sequential activity data. Extending upon this, we now propose a metric, rGP, which aims to measure the effectiveness of learned temporal patterns for a dataset and can be used as a decision metric whether to include recurrent layers into a network at all. Even for datasets including activities without explicit temporal processes, the rGP can be high, suggesting that temporal patterns were learned, and consequently convolutional networks are being outperformed by networks including recurrent layers. We conclude this article by putting forward the question to what degree popular HAR datasets contain unwanted temporal dependencies, which if not taken care of, can benefit networks in achieving high benchmark scores and give a false sense of overall generability to a real-world setting.

Ocean surface currents (OSC), abiotic features of the environment, are continuous and directed movements of ocean water. Prediction to OSC is of significant interests in physical oceanography. Recently, deep learning technology has shown feasibility in mining the intrinsic change pattern of marine numerical values, such as SSS, SST and SSHA, but not been applied to OSC prediction yet. In this work, a deep learning method, named skipped dual path network (SDPNet), is proposed for OSC prediction. Specifically, SDPNet has a convolutional neural network (CNN) module with a one-dimensional convolution layer, and a recurrent neural network (RNN) module with a dual-path structure. Each path consists of a stack of three layers of long short-term memory (LSTM) and gated recurrent unit (GRU), respectively. As well, a skipped-connection structure is added in the both paths. It aims to gradually mine the intrinsic change pattern contained in OSC time series data itself. Experiments are conducted on the South China Sea OSC data set in REDOS. SDPNet achieves accuracy 80.83%, 75.9%, 74.9%, 73.9%, 72.58%, 70.35%, 69.93% in predicting the coming 7 days OSC values and directions. It performs better than state-of-the-art machine learning methods, include Artificial Neural Network, Simple RNN, LSTM and GRU.

The study presents the results of the development and testing of deep learning neural network architectures, which demonstrate high accuracy rates in classifying neurophysiological data, in particular hemodynamic brain activation patterns obtained by functional near-infrared spectroscopy, during solving mathematical problems on spatial-numerical associations. The analyzed signal represents a multidimensional time series of oxyhemoglobin and deoxyhemoglobin dynamics. Taking the specificity of the fNIRS signal into account, a comparative analysis of 2 types of neural network architectures was carried out: (1) architectures based on recurrent neural networks: recurrent neural network with long short-term memory, recurrent neural network with long short-term memory with fully connected layers, bidirectional recurrent neural network with long short-term memory, convolutional recurrent neural network with long short-term memory; (2) architectures based on convolutional neural networks with 1D convolutions: convolutional neural network, fully convolutional neural network, residual neural network. Trained long short-term memory recurrent neural network architectures showed worse results in accuracy in comparison with 1D convolutional neural network architectures. Residual neural network (model_Resnet) demonstrated the highest accuracy rates in three experimental conditions more than 88% in detecting age-related differences in brain activation during spatial-numerical association tasks considering the individual characteristics of the respondents’ signal.

In recent years, supervised video summarization has achieved promising progress with various recurrent neural networks (RNNs) based methods, which treats video summarization as a sequence-to-sequence learning problem to exploit temporal dependency among video frames across variable ranges. However, RNN has limitations in modelling the long-term temporal dependency for summarizing videos with thousands of frames due to the restricted memory storage unit. Therefore, in this paper we propose a stacked memory network called SMN to explicitly model the long dependency among video frames so that redundancy could be minimized in the video summaries produced. Our proposed SMN consists of two key components: Long Short-Term Memory (LSTM) layer and memory layer, where each LSTM layer is augmented with an external memory layer. In particular, we stack multiple LSTM layers and memory layers hierarchically to integrate the learned representation from prior layers. By combining the hidden states of the LSTM layers and the read representations of the memory layers, our SMN is able to derive more accurate video summaries for individual video frames. Compared with the existing RNN based methods, our SMN is particularly good at capturing long temporal dependency among frames with few additional training parameters. Experimental results on two widely used public benchmark datasets: SumMe and TVsum, demonstrate that our proposed model is able to clearly outperform a number of state-of-the-art ones under various settings.

Human action recognition in videos is an important task with a broad range of applications. In this study, we improve the performance of recurrent attention convolutional neural network (RACNN) by proposing a novel model, “attention-again”. We consider the nature of video frames as sequences, which will cause the change of regions of interest in the frame, thus we cannot use an attention mechanism similar to that in images. “Attention-again” model is a variant from traditional attention model for recognizing human activities and is embedded in two long short-term memory (LSTM) layers. Different from hierarchal LSTM which change the LSTM structure to combine the hidden states from two LSTM layers, our proposals introduce “attention-again” model to avoid the change of LSTM structure. Furthermore, this model not only learns the relations in each frame, but also obtains the relations among all frames, and these relations instruct the next learning stage. Therefore, our proposed model outperform the baseline and is superior to methods with the same experimental conditions on three benchmark datasets: UCF-11, HMDB-51 and UCF-101. To understand how the model works, we also visualize the region of interest in the frame.

Recurrent neural networks (RNN) are at the core of modern automatic speech recognition (ASR) systems. In particular, long short-term memory (LSTM) recurrent neural networks have achieved state-of-the-art results in many speech recognition tasks, due to their efficient representation of long and short term dependencies in sequences of inter-dependent features. Nonetheless, internal dependencies within the element composing multidimensional features are weakly considered by traditional real-valued representations. We propose a novel quaternion long short-term memory (QL-STM) recurrent neural network that takes into account both the external relations between the features composing a sequence, and these internal latent structural dependencies with the quaternion algebra. QLSTMs are compared to LSTMs during a memory copy-task and a realistic application of speech recognition on the Wall Street Journal (WSJ) dataset. QLSTM reaches better performances during the two experiments with up to 2.8 times less learning parameters, leading to a more expressive representation of the information.

This paper develops a general framework for learning interpretable data representation via Long Short-Term Memory (LSTM) recurrent neural networks over hierarchal graph structures. Instead of learning LSTM models over the pre-fixed structures, we propose to further learn the intermediate interpretable multi-level graph structures in a progressive and stochastic way from data during the LSTM network optimization. We thus call this model the structure-evolving LSTM. In particular, starting with an initial element-level graph representation where each node is a small data element, the structure-evolving LSTM gradually evolves the multi-level graph representations by stochastically merging the graph nodes with high compatibilities along the stacked LSTM layers. In each LSTM layer, we estimate the compatibility of two connected nodes from their corresponding LSTM gate outputs, which is used to generate a merging probability. The candidate graph structures are accordingly generated where the nodes are grouped into cliques with their merging probabilities. We then produce the new graph structure with a Metropolis-Hasting algorithm, which alleviates the risk of getting stuck in local optimums by stochastic sampling with an acceptance probability. Once a graph structure is accepted, a higher-level graph is then constructed by taking the partitioned cliques as its nodes. During the evolving process, representation becomes more abstracted in higher-levels where redundant information is filtered out, allowing more efficient propagation of long-range data dependencies. We evaluate the effectiveness of structure-evolving LSTM in the application of semantic object parsing and demonstrate its advantage over state-of-the-art LSTM models on standard benchmarks.

This article explores the suitability of a long short-term memory recurrent neural network (LSTM-RNN) and artificial intelligence (AI) method for low-flow time series forecasting. The long short-term memory works on the sequential framework which considers all of the predecessor data. This forecasting method used daily discharged data collected from the Basantapur gauging station located on the Mahanadi River basin, India. Different metrics [root-mean-square error (RMSE), Nash–Sutcliffe efficiency (ENS), correlation coefficient (R) and mean absolute error] were selected to assess the performance of the model. Additionally, recurrent neural network (RNN) model is also used to compare the adaptability of LSTM-RNN over RNN and naive method. The results conclude that the LSTM-RNN model (R = 0.943, ENS = 0.878, RMSE = 0.487) outperformed RNN model (R = 0.935, ENS = 0.843, RMSE = 0.516) and naive method (R = 0.866, ENS = 0.704, RMSE = 0.793). The finding of this research concludes that LSTM-RNN can be used as new reliable AI technique for low-flow forecasting.

With the development of China's Belt and Road Initiative (BRI), the port plays a significant role and its operation management faces some pressure. In this regard, prediction of daily container volumes will provide the manager with data support for better plan of a storage yard. In this work, by deep learning the historical dataset, the long short-term memory (LSTM) recurrent neural network (RNN) is trained and used to predict daily volumes of containers which will enter the storage yard. The raw dataset of a certain port from 2013 to 2016 is chosen as the training set and the dataset of 2017 is used as the test set to evaluate the performance of the proposed prediction model. Then the LSTM model is established with Python and Tensorflow framework. The structure parameters are adjusted to find the optimal LSTM network, so as to improve the prediction accuracy. It appears that the LSTM model with two hidden layers and 30 hidden layer units has less prediction error between the real data and predicted data of 2017. The prediction error of daily container volumes between predicted value and real data of 2017 is about 12.39%, which is less than the people-predicted error. It is promising that the proposed LSTM RNN model can be applied to predict the daily volumes of containers and have higher prediction accuracy.

In this paper, we develop a method based on a deep learning method in financial market prediction, which includes BRICS economies as the test cases. Financial markets are rife with volatility that is affected by a "bed of complexity," coddled by local and distal factors. To leverage these vast datasets both deep learning models such as Convolutional Neural Networks (CNNs), Long Short Term Memory (LSTM) networks as well as hybrid architectures are used in this study. The paper evaluates the predictive accuracy of the models, and by so doing, identifies their strengths in predicting temporal dependencies and intricate market patterns. In particular, deep learning techniques are applied to case studies of individual countries in the BRICS to highlight the application of deep learning to disparate country specific problems, such as liquidity crises and market shocks. These findings show that classical statistical methods are outperformed by deep learning systems in a precise and reliable financial forecasting. This research highlights the ability of AI driven systems to change financial decision making processes, improving investor confidence and improving economic stability in BRICS nations. This study also readers the value of deep learning in financial market analysis, especially in economies in the developing countries. Application of techniques and architectures e.g. Convolutional Neural Networks (CNNs) that excel at identifying spatial patterns, and Long Short-Term Memory (LSTM) networks renowned for their prowess on sequential and time series data, for real world market prediction are explained. In addition, the research discusses hybrid architectures which extend knowledge, fusing strengths of both architectures to improve prediction accuracy and how deep learning develops to solve particular financial challenges. Through reading these notes readers get exposed to data preprocessing techniques such as normalization and feature selection which are important for boosting deep learning performance. The paper also includes an introduction to the evaluation of models using MSE and R-squared values for validating them in terms of reliable outputs. This research combines deep learning theory and practical case study to offer a useful educational resource for students, researchers, and practitioners who want to apply AI in financial forecasting in complex and dynamic global markets.

This paper proposes a multi-layered anomaly detection scheme to train feature extraction and to test anomaly prediction by using Convolutional Neural Networks (CNNs) layer, Bidirectional and Unidirectional Long Short-Term Memory (LSTM) Recurrent Neural Networks (RNNs), which is one of a novel deep architecture named stacked convolutional bidirectional LSTM network (SCB-LSTM). In the proposed model, the stacked CNNs perform feature extraction of vibration sensor signal patterns, and the result is used to feature learning with the stacked bidirectional LSTMs (SB-LSTMs). After this procedure, the stacked unidirectional LSTMs (SU-LSTMs) enhance the feature learning, and a regression layer finally predicts anomaly detections. The experimental results of bearing data not only show the accuracy of the proposed model in anomaly detection for rotating machinery diagnostics, but also suggest the better performance than other state-of-the-art algorithms such as a plain uni-LSTM or Bi-LSTM.

Maxout neurons for deep convolutional and LSTM neural networks in speech recognition

Artificial intelligence models have been widely applied for natural gas consumption forecasting over the past decades, especially for short-term consumption forecasting. This paper proposes a three-layer neural network forecasting model that can extract key information from input factors and improve the weight optimization mechanism of long short-term memory (LSTM) neural network to effectively forecast short-term consumption. In the proposed model, a convolutional neural network (CNN) layer is adopted to extract the features among various factors affecting natural gas consumption and improve computing efficiency. The LSTM layer is able to learn and save the long-distance state through the gating mechanism and overcomes the defects of gradient disappearance and explosion in the recurrent neural network. To solve the problem of encoding input sequences as fixed-length vectors, the layer of attention (ATT) is used to optimize the assignment of weights and highlight the key sequences. Apart from the comparisons with other popular forecasting models, the performance and robustness of the proposed model are validated on datasets with different fluctuations and complexities. Compared with traditional two-layer models (CNN-LSTM and LSTM-ATT), the mean absolute range normalized errors (MARNE) of the proposed model in Athens and Spata are improved by more than 16% and 11%, respectively. In comparison with single LSTM, back propagation neural network, support vector regression, and multiple linear regression methods, the improvement in MARNE exceeds 42% in Athens. The coefficient of determination is improved by more than 25%, even in the high-complexity dataset, Spata.

Autism Spectrum Disorder (ASD) is a heterogeneous disorder in children, and the current clinical diagnosis is accomplished using behavioral, cognitive, developmental, and language metrics. These clinical metrics can be imperfect measures as they are subject to high test-retest variability, and are influenced by assessment factors such as environment, social structure, or comorbid disorders. Advances in neuroimaging coupled with machine-learning provides an opportunity to develop methods that are more quantifiable, and reliable than existing clinical techniques. In this paper, we design and develop a deep-learning model that operates on functional magnetic resonance imaging (fMRI) data, and can classify between ASD and neurotypical brains. We introduce a novel strategy to transform time-series data extracted from fMRI signals into Gramian Angular Field (GAF) while locking in the temporal and spatial patterns in the data. Our motivation is to design and develop a novel framework that could encode the time-series, acquired from fMRI data, into images that can be used by deep-learning architectures that have been successful in computer vision. In our proposed framework called ASD-GResTM, we used a Convolutional Neural Network (CNN) to extract useful features from GAF images. We then used a Long Short-Term Memory (LSTM) layer to learn the activities between the regions. Finally, the output representations of the last LSTM layer are applied to a single-layer perceptron (SPL) to get the final classification. Our extensive experimentation demonstrates high accuracy across 4 centers, and outperforms state-of-the-art models on two centers with an increase in the accuracy of 17.58% and 6.7%, respectively as compared to the state of the art. Our model achieved the maximum accuracy of 81.78% with high degree of sensitivity and specificity. All training, validation, and testing was accomplished using openly available ABIDE-I benchmarking dataset.

With the huge success of applying deep learning (DL) methodologies to image recognition and natural language processing in recent years, researchers are now keen to use them in the machine condition monitoring (MCM) context. There are numerous papers in applying various DL techniques, such as auto-encoder, restricted Boltzmann machine, convolutional neural network and recurrent neural network, etc., to MCM problems ranging from component-level condition monitoring (machine tool wear prediction, bearing fault diagnosis and classification and hydraulic pump fault diagnosis) to system-level health management (aircraft and spacecraft diagnosis). In this paper, we give a brief overview in the area of DL for MCM with a focus on reviewing the most recent papers published since 2019. In Part 1, we present some critical views regarding whether any breakthrough has been achieved from an MCM domain expert perspective, with the main conclusion that DL has great potential for MCM applications, and a major breakthrough could come soon since the shortfalls lie more in data than in the DL methodologies. Our overall impression is that (a) DL models are not really showing their great potentials with only a small training data; (b) faulty-condition data is hard to come by for training DL, but normal condition data is abundant, so anomaly detection makes more sense; (c) applying DL only to the Case Western Reserve University (CWRU) bearing fault dataset is not sufficient for real world industrial applications as it was from a very simple test rig, and applying DL to data from complex systems like helicopter gearbox data may deliver much more convincing results. In Part 2, we enhance the main conclusion of the critical review with supplement views and a case study on analysing Bell-206B helicopter main gearbox planet bearing failure data using some traditional MCM techniques in contrast to applying the long short-term memory (LSTM) DL method. We can conclude from the case study that the DL-based methods are not necessarily always superior to the traditional MCM techniques for dataset from moderately complex machinery.