Data Augmentation of IMU Signals and Evaluation via a Semi-Supervised Classification of Driving Behavior

• URL: http://arxiv.org/abs/2006.09267v1
• Date: Tue, 16 Jun 2020 15:49:21 GMT
• Title: Data Augmentation of IMU Signals and Evaluation via a Semi-Supervised Classification of Driving Behavior
• Authors: Amani Jaafer and Gustav Nilsson and Giacomo Como
• Abstract summary: We present a semi-supervised learning solution to classify portions of trips according to whether drivers are driving aggressively or normally. Our results show that, by utilizing RCGAN-generated labeled data, the classification of the drivers is improved in 79% of the cases.
• Score: 4.640835690336653
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Over the past years, interest in classifying drivers' behavior from data has surged. Such interest is particularly relevant for car insurance companies who, due to privacy constraints, often only have access to data from Inertial Measurement Units (IMU) or similar. In this paper, we present a semi-supervised learning solution to classify portions of trips according to whether drivers are driving aggressively or normally based on such IMU data. Since the amount of labeled IMU data is limited and costly to generate, we utilize Recurrent Conditional Generative Adversarial Networks (RCGAN) to generate more labeled data. Our results show that, by utilizing RCGAN-generated labeled data, the classification of the drivers is improved in 79% of the cases, compared to when the drivers are classified with no generated data.

Related papers

• Dynamic Data Pruning for Automatic Speech Recognition [58.95758272440217]
  We introduce Dynamic Data Pruning for ASR (DDP-ASR), which offers fine-grained pruning granularities specifically tailored for speech-related datasets. Our experiments show that DDP-ASR can save up to 1.6x training time with negligible performance loss.
  arXiv  Detail & Related papers  (2024-06-26T14:17:36Z)
• A Bayesian Approach for Prioritising Driving Behaviour Investigations in Telematic Auto Insurance Policies [0.6249768559720121]
  We make use of trip GPS and accelerometer data, augmented by geospatial information, to train an imperfect classifier for delivery driving on a per-trip basis. A posterior probability is converted to a priority score, which was used to select the most valuable candidates for manual investigation. The approach has achieved a significant improvement in efficiency of human resource allocation compared to manual searching.
  arXiv  Detail & Related papers  (2024-04-22T15:26:24Z)
• D2E-An Autonomous Decision-making Dataset involving Driver States and Human Evaluation [6.890077875318333]
  Driver to Evaluation dataset (D2E) is an autonomous decision-making dataset. It contains data on driver states, vehicle states, environmental situations, and evaluation scores from human reviewers. D2E contains over 1100 segments of interactive driving case data covering from human driver factor to evaluation results.
  arXiv  Detail & Related papers  (2024-04-12T21:29:18Z)
• Exploiting T-norms for Deep Learning in Autonomous Driving [60.205021207641174]
  We show how it is possible to define memory-efficient t-norm-based losses, allowing for exploiting t-norms for the task of event detection in autonomous driving.
  arXiv  Detail & Related papers  (2024-02-17T18:51:21Z)
• G-MEMP: Gaze-Enhanced Multimodal Ego-Motion Prediction in Driving [71.9040410238973]
  We focus on inferring the ego trajectory of a driver's vehicle using their gaze data. Next, we develop G-MEMP, a novel multimodal ego-trajectory prediction network that combines GPS and video input with gaze data. The results show that G-MEMP significantly outperforms state-of-the-art methods in both benchmarks.
  arXiv  Detail & Related papers  (2023-12-13T23:06:30Z)
• Anomalous Behavior Detection in Trajectory Data of Older Drivers [0.8426358786287627]
  We propose an Edge-Attributed Matrix that can represent the key properties of temporally-detailed trajectory datasets. Experiments using real-world datasets demonstrated that our approach identifies abnormal driving behaviors.
  arXiv  Detail & Related papers  (2023-11-29T17:22:28Z)
• Augmentation-Aware Self-Supervision for Data-Efficient GAN Training [68.81471633374393]
  Training generative adversarial networks (GANs) with limited data is challenging because the discriminator is prone to overfitting. We propose a novel augmentation-aware self-supervised discriminator that predicts the augmentation parameter of the augmented data. We compare our method with state-of-the-art (SOTA) methods using the class-conditional BigGAN and unconditional StyleGAN2 architectures.
  arXiv  Detail & Related papers  (2022-05-31T10:35:55Z)
• Unsupervised Driving Behavior Analysis using Representation Learning and Exploiting Group-based Training [15.355045011160804]
  Driving behavior monitoring plays a crucial role in managing road safety and decreasing the risk of traffic accidents. Current work performs a robust driving pattern analysis by capturing variations in driving patterns. It forms consistent groups by learning compressed representation of time series.
  arXiv  Detail & Related papers  (2022-05-12T10:27:47Z)
• Driving Anomaly Detection Using Conditional Generative Adversarial Network [26.45460503638333]
  This study proposes an unsupervised method to quantify driving anomalies using a conditional generative adversarial network (GAN) The approach predicts upcoming driving scenarios by conditioning the models on the previously observed signals. The results are validated with perceptual evaluations, where annotators are asked to assess the risk and familiarity of the videos detected with high anomaly scores.
  arXiv  Detail & Related papers  (2022-03-15T22:10:01Z)
• One Million Scenes for Autonomous Driving: ONCE Dataset [91.94189514073354]
  We introduce the ONCE dataset for 3D object detection in the autonomous driving scenario. The data is selected from 144 driving hours, which is 20x longer than the largest 3D autonomous driving dataset available. We reproduce and evaluate a variety of self-supervised and semi-supervised methods on the ONCE dataset.
  arXiv  Detail & Related papers  (2021-06-21T12:28:08Z)
• Classify and Generate Reciprocally: Simultaneous Positive-Unlabelled Learning and Conditional Generation with Extra Data [77.31213472792088]
  The scarcity of class-labeled data is a ubiquitous bottleneck in many machine learning problems. We address this problem by leveraging Positive-Unlabeled(PU) classification and the conditional generation with extra unlabeled data. We present a novel training framework to jointly target both PU classification and conditional generation when exposed to extra data.
  arXiv  Detail & Related papers  (2020-06-14T08:27:40Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.