A study on attention-based LSTM for abnormal behavior recognition with variable pooling.

A social attention module is used to aggregate neighbors' interactions based on the correlation between pedestrians' moving directions and future trajectories. ... Generative model-based methods handle future uncertainties by sampling a latent variable. However, few studies explored the generation of the latent variable. ... Social-LSTM [6]: An improved LSTM-based trajectory for multi-future trajectory predictions based on conditional prediction method by proposing a social pooling layer to variational recurrent ...

arXiv:2002.01852v3 fatcat:45g6rucccrcmrd6qtbjrshrng4

Multiple Versions

In this thesis, our focus is on the state-ofthe-art methods for human behavior recognition based on deep learning. ... 2) The YOLOv3 + LSTM network to reply on both spatiotemporal information with class score fusion is able to achieve 97.58% accuracy based on our dataset for sign language processing. ... A low-rank approximation of second order pooling (attentional pooling) is to replace mean pooling or max pooling in the last pooling layer of CNN network so as to achieve the behavior recognition . ...

doi:10.1109/ivcnz51579.2020.9290640 fatcat:sq4fni6z2nfz5okecnsbmzum6e

To address and explore the information of users' behaviors in continuous time spaces, we propose to use time attention based recurrent layers to embed the detailed information of the time interval, such ... In this work, we present a novel method for detecting fraud transactions by leveraging patterns from both users' static profiles and users' dynamic behaviors in a unified framework. ... . • Self-attention LSTM: We add a self attention layer on the top of Bi-LSTM which is introduced in [12] . • CNN+Max pooling: We use traditional CNN with Max pooling to extract click and transaction behavior's ...

arXiv:1912.11760v2 fatcat:6ksgrgsh2bafxk62jzyjash3g4

Multiple Versions

In this paper, we focus solely on data fusion and recognition techniques in the context of vision with an RGB-D perspective. ... This review is aimed to scope current literature on data fusion and action recognition techniques and to identify gaps and future research direction. ... For mapping video frames with variable length inputs to variable length outputs, Donahue et al. [120] have proposed an LRCN. ...

doi:10.3390/s21124246 fatcat:7dvocdy63rckne5yunhfsnr4p4

DOAJ

Various research efforts have approached the problem of detecting abnormal human driver behavior with the aid of capturing and analyzing the face of driver and vehicle dynamics via image and video processing ... However, with the advent of deep learning algorithms, a significant amount of research has also been conducted to predict and analyze driver's behavior or action related information using neural network ... Bidirectional LSTM with attention mechanism combines birdirectional LSTM with attention network [94] . C. ...

doi:10.1109/access.2020.2999829 fatcat:5nxtzm6yfbe4jf6nqgreqw45r4

DOAJ

At present, the existing abnormal event detection models based on deep learning mainly focus on data represented by a vectorial form, which pay little attention to the impact of the internal structure ... Second, we project expectation subgraphs on the sparse vector to combine with a support vector classifier to calibrate the results of a linear support vector classifier. ... For example, the study in [20] presents a complex event processing method based on trajectories. Song et al. ...

doi:10.1109/access.2020.2997357 fatcat:yrytr47smnapdla7cwo7sbyqma

DOAJ

In this paper, we propose a new method for automatic classification of arrhythmias based on deep neural networks (DNNs). ... The algorithm is evaluated based on the test set of China Physiological Signal Challenge (CPSC) dataset with F 1 measure regarded as the harmonic mean between the precision and recall. ... LSTM layer stacked with a global maximum pooling (GMP) layer. ...

doi:10.1109/access.2019.2931500 fatcat:wosycpspivai3cjkrjuwltnyna

DOAJ

In crowd scenarios, predicting trajectories of pedestrians is a complex and challenging task depending on many external factors. ... This paper compares these relatively new deep learning algorithms with classical knowledge-based models that are widely used to simulate pedestrian dynamics. ... [40] the generator is composed of an LSTM-based encoder, a context pooling module, and an LSTM-based decoder. The discriminator uses LSTMs as well. ...

arXiv:2111.06740v1 fatcat:lstxi2qrhrhdtlpm47x4vpinry

Open Access

Besides, a body-region attention and a global-part feature pooling strategy are devised for individual action recognition. ... Moreover, we adopt a spatio-temporal attention model to focus on key persons/frames for improved recognition performance. ... INTRODUCTION U NDERSTANDING dynamic scenes in sports games and surveillance videos encompasses a wide range of applications, like sports team tactics analysis and abnormal behavior detection. ...

doi:10.1109/tcsvt.2019.2894161 fatcat:wcjvyo3wgfbsfcew4x62sw6cfi

A conditional variational auto-encoder module is trained to learn the latent space of possible future paths based on attentive dynamic maps for interaction modeling and then is used to predict multiple ... How an agent moves is affected by the various behaviors of its neighboring agents in different environments. ... [12] is a rule-based model with the repulsive force for collision avoidance and the attractive force for social connections; • Social LSTM [8] proposes Social pooling with a rectangular occupancy ...

arXiv:2006.08264v2 fatcat:ka2yhsomfbaz3omt3jngfymk2u

Multiple Versions

Then, we construct a Memory Network consisting of linear (Autoregressive Model) and non-linear predictions (Bidirectional LSTM with Attention) to capture temporal dependence from time-series data. ... We first build a Deep Convolutional Autoencoder to characterize spatial dependence of multi-sensor data with a Maximum Mean Discrepancy (MMD) to better distinguish between the noisy, normal, and abnormal ... Besides, some learning algorithms based on time-series data have been studied for decades. ...

arXiv:2107.12626v2 fatcat:t6eyfskylnfhdhkv6btrzxtugu

Open Access Multiple Versions

To achieve this goal, a vehicle lane-changing prediction model based on game theory and deep learning is developed. ... For the deep-learning component, long short-term memory and a convolutional neural network are used to extract and learn data features during the lane-changing process as well as combine the output of ... each other. is allows researchers to conduct more in-depth and accurate studies on driving behavior, such as those on the recognition of drunk driving, recognition of fatigue driving, and classification ...

doi:10.1155/2021/6634960 fatcat:7w35y74ojrdhtcwlcgb3olj7ii

DOAJ

This paper presents a novel and effective audio based method on depression classification. ... Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) to deliver a more comprehensive audio representation. ... In DepAudioNet, we apply 128 convolution maps being of 3 × 1 on the time-frequency of 2D representation capturing the variability of 64ms precision, with the max-pooling operation on a region being size ...

doi:10.1145/2988257.2988267 dblp:conf/mm/MaYC0W16 fatcat:sn4k2isxnjfynpr7cfwtwgmdne

The review is concluded with a debate on what challenges need to be overcome in the domain in the near future. ... A survey on existing systems for recognizing emotions in real-life scenarios—their possibilities, limitations, and identified problems—is also provided. ... Zhao et al. developed a model that combines attention-based bidirectional LSTM with attention-based fully convolutional networks to obtain spatial-temporal features from the speech spectrogram [55] . ...

doi:10.3390/electronics11030496 fatcat:pagrnyshp5fq7nkcdcqd2gzdbm

DOAJ

INDEX TERMS Artificial intelligence, artificial neural networks, risk analysis, attention mechanism, CNN, RNN, driving behavior recognition, smartphone sensor data. ... Driving behavior recognition is a challenging task that exploits the acceleration and angular velocity information of the vehicle collected by smartphone to identify various driving events. ... METHOD OF BUILDING DRIVING BEHAVIOR DATASET At presently, there are a few datasets for driving behavior recognition based on smartphone sensor data. ...

doi:10.1109/access.2019.2932434 fatcat:i3yjf3fvcbh45f3wv7z3b7sc4a

DOAJ

TPPO: A Novel Trajectory Predictor with Pseudo Oracle [article]

Preserved Fulltext

Other Versions

Deep Learning Methods for Human Behavior Recognition

Preserved Fulltext

A Time Attention based Fraud Transaction Detection Framework [article]

Preserved Fulltext

Other Versions

RGB-D Data-Based Action Recognition: A Review

Preserved Fulltext

Detecting Human Driver Inattentive and Aggressive Driving Behavior using Deep Learning: Recent Advances, Requirements and Open Challenges

Preserved Fulltext

Abnormal Event Detection via Feature Expectation Subgraph Calibrating Classification in Video Surveillance Scenes

Preserved Fulltext

Automatic cardiac arrhythmia classification using combination of deep residual network and bidirectional LSTM

Preserved Fulltext

Review of Pedestrian Trajectory Prediction Methods: Comparing Deep Learning and Knowledge-based Approaches [article]

Preserved Fulltext

stagNet: An Attentive Semantic RNN for Group Activity and Individual Action Recognition

Preserved Fulltext

AMENet: Attentive Maps Encoder Network for Trajectory Prediction [article]

Preserved Fulltext

Other Versions

Unsupervised Deep Anomaly Detection for Multi-Sensor Time-Series Signals [article]

Preserved Fulltext

Other Versions

Lane-Changing Behavior Prediction Based on Game Theory and Deep Learning

Preserved Fulltext

DepAudioNet

Preserved Fulltext

Bringing Emotion Recognition Out of the Lab into Real Life: Recent Advances in Sensors and Machine Learning

Preserved Fulltext

Attention-Based Convolutional and Recurrent Neural Networks for Driving Behavior Recognition Using Smartphone Sensor Data

Preserved Fulltext