Trajectory prediction is a crucial challenge in autonomous vehicle motion planning and decision-making techniques. However, existing methods face limitations in accurately capturing vehicle dynamics and interactions. ...
详细信息
Trajectory prediction is a crucial challenge in autonomous vehicle motion planning and decision-making techniques. However, existing methods face limitations in accurately capturing vehicle dynamics and interactions. To address this issue, this paper proposes a novel approach to extracting vehicle velocity and acceleration, enabling the learning of vehicle dynamics and encoding them as auxiliary information. The VDI-LSTM model is designed, incorporating graph convolution and attention mechanisms to capture vehicle interactions using trajectory data and dynamic information. Specifically, a dynamics encoder is designed to capture the dynamic information, a dynamic graph is employed to represent vehicle interactions, and an attention mechanism is introduced to enhance the performance of LSTM and graph convolution. To demonstrate the effectiveness of our model, extensive experiments are conducted, including comparisons with several baselines and ablation studies on real-world highway datasets. Experimental results show that VDI-LSTM outperforms other baselines compared, which obtains a 3% improvement on the average RMSE indicator over the five prediction steps.
Multimodal sentiment analysis has become a popular research topic in recent years. However, existing methods have two unaddressed limitations: (1) they use limited supervised labels to train models, which makes it imp...
详细信息
Multimodal sentiment analysis has become a popular research topic in recent years. However, existing methods have two unaddressed limitations: (1) they use limited supervised labels to train models, which makes it impossible for model to fully learn sentiments in different modal data; (2) they employ text and image pre-trained models trained in different unimodal tasks to extract different modal features, so that the extracted features cannot take into account the interactive information between image and text. To solve these problems, in this paper we propose a Vision-Language Contrastive Learning network (VLCLNet). First, we introduce a pre-trained Large Language Model (LLM), which is trained from vast quantities of multimodal data, has better understanding ability for image and text contents, thus being effectively applied to different tasks while requiring few amount of labelled training data. Second, we adapt a Multimodal Large Language Model (MLLM), BLIP-2 (Bootstrapping Language-Image Pre-training) network, to extract multimodal fusion feature. Such MLLM can fully consider the correlation between images and texts when extracting features. In addition, due to the discrepancy between the pre-training task and the sentiment analysis task, the pre-trained model will output the suboptimal prediction results. We use LoRA (Low-Rank Adaptation) fine-tuning strategy to update the model parameters on sentiment analysis task, which avoids the issue of inconsistent task between pre-training task and downstream task. Experiments verify that the proposed VLCLNet is superior to other strong baselines.
The twentieth century ended with the vision of smart dust: a network of wirelessly connected devices whose size would match that of a dust particle, each one a se- containedpackageequippedwithsensing,computation,commu...
详细信息
ISBN:
(数字)9781441958341
ISBN:
(纸本)9781441958334;9781489990587
The twentieth century ended with the vision of smart dust: a network of wirelessly connected devices whose size would match that of a dust particle, each one a se- containedpackageequippedwithsensing,computation,communication,andpower. Smart dust held the promise to bridge the physical and digital worlds in the most unobtrusive manner, blending together realms that were previously considered well separated. Applications involved scattering hundreds, or even thousands, of smart dust devices to monitor various environmental quantities in scenarios ranging from habitat monitoring to disaster management. The devices were envisioned to se- organize to accomplish their task in the most ef?cient way. As such, smart dust would become a powerful tool, assisting the daily activities of scientists and en- neers in a wide range of disparate disciplines. Wireless sensor networks (WSNs), as we know them today, are the most no- worthy attempt at implementing the smart dust vision. In the last decade, this ?eld has seen a fast-growing investment from both academia and industry. Signi?cant ?nancial resources and manpower have gone into making the smart dust vision a reality through WSNs. Yet, we still cannot claim complete success. At present, only specialist computerscientists or computerengineershave the necessary background to walk the road from conception to a ?nal, deployed, and running WSN system.
This book constitutes the refereed proceedings of the 18th Conference of the Spanish Association for;artificialintelligence, CAEPIA 2018, held in Granada, Spain, in October 2018.;The 36 full papers presented w...
详细信息
ISBN:
(数字)9783030003746
ISBN:
(纸本)9783030003739
This book constitutes the refereed proceedings of the 18th Conference of the Spanish Association for;artificialintelligence, CAEPIA 2018, held in Granada, Spain, in October 2018.;The 36 full papers presented were carefully selected from 240 submissions. The Conference of the Spanish Association of;artificialintelligence (CAEPIA) is a biennial forum open to researchers from all over the world to present and discuss their latest scientific and technological advances in Antificial intelligence (AI). Authors are kindly requested to submit unpublished original papers describing relevant research on AI issues from all points of view: formal, methodological, technical or applied.
Few-shot learning (FSL) aims to classify a novel object into a specific category under limited training samples. This is a challenging task since (1) the features expressed by pre-trained knowledge introduce perceived...
详细信息
Few-shot learning (FSL) aims to classify a novel object into a specific category under limited training samples. This is a challenging task since (1) the features expressed by pre-trained knowledge introduce perceived bias and then constrain the classification space, and (2) the use of general hallucination techniques based on global features fails to escape the limited classification space, resulting in suboptimal improvements. To solve these issues, this paper proposes an interventional feature generation (IFG) method. Specifically, we first use the relations of the categories or instances as interventional operations to implicitly constrain the feature representations (pre-trained knowledge) into different classification subsets. Then, we employ a parameter-free feature generation strategy to enrich each subset’s training samples of the support category. In other words, IFG provides a multi-subsets learning strategy to reduce the influence of perceived bias, enrich the diversity of generated features, and improve the robustness of the few-shot classifier. We apply our method to four benchmark datasets and observe state-of-the-art performance across all experiments. Specifically, compared to the baseline on the Mini-ImageNet dataset, our approach yields accuracy improvements of 6.03% and 3.46% for 1 and 5 support training samples, respectively. Furthermore, the proposed interventional feature generation technique can improve classifier performance in other FSL methods, demonstrating its versatility and potential for broader applications. The code is available at https://***/ShuoWangCS/IFG-FSL/.
This book features high-quality research papers presented at the 6th International Conference on Computational intelligence in Pattern Recognition (CIPR 2024), held at Maharaja Sriram Chandra Bhanja Deo University (MS...
详细信息
ISBN:
(数字)9789819780907
ISBN:
(纸本)9789819780891
This book features high-quality research papers presented at the 6th International Conference on Computational intelligence in Pattern Recognition (CIPR 2024), held at Maharaja Sriram Chandra Bhanja Deo University (MSCB University), Baripada, Odisha, India, during March 15–16, 2024. It includes practical development experiences in various areas of data analysis and pattern recognition, focusing on soft computing technologies, clustering and classification algorithms, rough set and fuzzy set theory, evolutionary computations, neural science and neural network systems, image processing, combinatorial pattern matching, social network analysis, audio and video data analysis, data mining in dynamic environments, bioinformatics, hybrid computing, big data analytics, and deep learning. It also provides innovative solutions to the challenges in these areas and discusses recent developments.
WSC2008Chair’s Welcome Message Dear Colleague, The World Soft Computing (WSC) conference is an annual international online conference on applied and theoretical soft computing technology. This WSC 2008 is the thirtee...
详细信息
ISBN:
(数字)9783540896197
ISBN:
(纸本)9783540896180
WSC2008Chair’s Welcome Message Dear Colleague, The World Soft Computing (WSC) conference is an annual international online conference on applied and theoretical soft computing technology. This WSC 2008 is the thirteenth conference in this series and it has been a great success. We received a lot of excellent paper submissions which were peer-reviewed by an international team of experts. Only60 papers out of111 submissions were selected for online publication. This assured a high quality standard for this online conference. The corresponding online statistics are a proof of the great world-wide interest in the WSC 2008 conference. The conference website had a total of33,367di?erent human user accessesfrom43 countries with around100 visitors every day,151 people signed up to WSC to discuss their scienti?c disciplines in our chat rooms and the forum. Also audio and slide presentations allowed a detailed discussion of the papers. The submissions and discussions showed that there is a wide range of soft computing applications to date. The topics covered by the conference range from applied to theoretical aspects of fuzzy, neuro-fuzzy and rough sets over to neural networks to single and multi-objective optimisation. Contributions aboutparticleswarmoptimisation,geneexpressionprogramming,clustering, classi?cation,supportvectormachines,quantumevolutionandagentsystems have also been received. One whole session was devoted to soft computing techniques in computer graphics, imaging, vision and signal processing.
暂无评论