Abstract
Global financial crime activity is driving demand for machine learning solutions in fraud prevention. However, prevention systems are commonly serviced to financial institutions in isolation, and few provisions exist for data sharing due to fears of unintentional leaks and adversarial attacks. Collaborative learning advances in finance are rare, and it is hard to find real-world insights derived from privacy-preserving data processing systems. In this paper, we present a collaborative deep learning framework for fraud prevention, designed from a privacy standpoint, and awarded at the recent PETs Prize Challenges. We leverage latent embedded representations of varied-length transaction sequences, along with local differential privacy, in order to construct a data release mechanism which can securely inform externally hosted fraud and anomaly detection models. We assess our contribution on two distributed data sets donated by large payment networks, and demonstrate robustness to popular inference-time attacks, along with utility-privacy trade-offs analogous to published work in alternative application domains.
A comprehensive survey of research towards AI-enabled unmanned aerial systems in pre-, active-, and post-wildfire management
Authors: Authors: Sayed Pedram Haeri Boroujeni, Abolfazl Razi, Sahand Khoshdel, Fatemeh Afghah, Janice L. Coen, Leo ONeill, Peter Z. Fule, Adam Watts, Nick-Marios T. Kokolakis, Kyriakos G. Vamvoudakis
Abstract
Wildfires have emerged as one of the most destructive natural disasters worldwide, causing catastrophic losses in both human lives and forest wildlife. Recently, the use of Artificial Intelligence (AI) in wildfires, propelled by the integration of Unmanned Aerial Vehicles (UAVs) and deep learning models, has created an unprecedented momentum to implement and develop more effective wildfire management. Although some of the existing survey papers have explored various learning-based approaches, a comprehensive review emphasizing the application of AI-enabled UAV systems and their subsequent impact on multi-stage wildfire management is notably lacking. This survey aims to bridge these gaps by offering a systematic review of the recent state-of-the-art technologies, highlighting the advancements of UAV systems and AI models from pre-fire, through the active-fire stage, to post-fire management. To this aim, we provide an extensive analysis of the existing remote sensing systems with a particular focus on the UAV advancements, device specifications, and sensor technologies relevant to wildfire management. We also examine the pre-fire and post-fire management approaches, including fuel monitoring, prevention strategies, as well as evacuation planning, damage assessment, and operation strategies. Additionally, we review and summarize a wide range of computer vision techniques in active-fire management, with an emphasis on Machine Learning (ML), Reinforcement Learning (RL), and Deep Learning (DL) algorithms for wildfire classification, segmentation, detection, and monitoring tasks. Ultimately, we underscore the substantial advancement in wildfire modeling through the integration of cutting-edge AI techniques and UAV-based data, providing novel insights and enhanced predictive capabilities to understand dynamic wildfire behavior.
Image-based Deep Learning for Smart Digital Twins: a Review
Authors: Authors: Md Ruman Islam, Mahadevan Subramaniam, Pei-Chi Huang (Department of Computer Science, University of Nebraska at Omaha, Omaha, NE, USA)
Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Systems and Control (eess.SY)
Abstract
Smart Digital twins (SDTs) are being increasingly used to virtually replicate and predict the behaviors of complex physical systems through continual data assimilation enabling the optimization of the performance of these systems by controlling the actions of systems. Recently, deep learning (DL) models have significantly enhanced the capabilities of SDTs, particularly for tasks such as predictive maintenance, anomaly detection, and optimization. In many domains, including medicine, engineering, and education, SDTs use image data (image-based SDTs) to observe and learn system behaviors and control their behaviors. This paper focuses on various approaches and associated challenges in developing image-based SDTs by continually assimilating image data from physical systems. The paper also discusses the challenges involved in designing and implementing DL models for SDTs, including data acquisition, processing, and interpretation. In addition, insights into the future directions and opportunities for developing new image-based DL approaches to develop robust SDTs are provided. This includes the potential for using generative models for data augmentation, developing multi-modal DL models, and exploring the integration of DL with other technologies, including 5G, edge computing, and IoT. In this paper, we describe the image-based SDTs, which enable broader adoption of the digital twin DT paradigms across a broad spectrum of areas and the development of new methods to improve the abilities of SDTs in replicating, predicting, and optimizing the behavior of complex systems.
Autonomous Multi-Rotor UAVs: A Holistic Approach to Design, Optimization, and Fabrication
Authors: Authors: Aniruth A, Chirag Satpathy, Jothika K, Nitteesh M, Gokulraj M, Venkatram K, Harshith G, Shristi S, Anushka Vani, Jonathan Spurgeon
Abstract
Unmanned Aerial Vehicles (UAVs) have become pivotal in domains spanning military, agriculture, surveillance, and logistics, revolutionizing data collection and environmental interaction. With the advancement in drone technology, there is a compelling need to develop a holistic methodology for designing UAVs. This research focuses on establishing a procedure encompassing conceptual design, use of composite materials, weight optimization, stability analysis, avionics integration, advanced manufacturing, and incorporation of autonomous payload delivery through object detection models tailored to satisfy specific applications while maintaining cost efficiency. The study conducts a comparative assessment of potential composite materials and various quadcopter frame configurations. The novel features include a payload-dropping mechanism, a unibody arm fixture, and the utilization of carbon-fibre-balsa composites. A quadcopter is designed and analyzed using the proposed methodology, followed by its fabrication using additive manufacturing and vacuum bagging techniques. A computer vision-based deep learning model enables precise delivery of payloads by autonomously detecting targets.
A Community Detection and Graph Neural Network Based Link Prediction Approach for Scientific Literature
Abstract
This study introduces an innovative approach that integrates community detection algorithms with Graph Neural Network (GNN) models to enhance link prediction in scientific literature networks. We specifically focus on the utilization of the Louvain community detection algorithm to uncover latent community structures within these networks, which are then incorporated into GNN architectures to predict potential links. Our methodology demonstrates the importance of understanding community dynamics in complex networks and leverages the strengths of both community detection and GNNs to improve predictive accuracy. Through extensive experiments on bipartite graphs representing scientific collaborations and citations, our approach not only highlights the synergy between community detection and GNNs but also addresses some of the prevalent challenges in link prediction, such as scalability and resolution limits. The results suggest that incorporating community-level information can significantly enhance the performance of GNNs in link prediction tasks. This work contributes to the evolving field of network science by offering a novel perspective on integrating advanced machine learning techniques with traditional network analysis methods to better understand and predict the intricate patterns of scientific collaborations.
Exploiting Polarized Material Cues for Robust Car Detection
Authors: Authors: Wen Dong, Haiyang Mei, Ziqi Wei, Ao Jin, Sen Qiu, Qiang Zhang, Xin Yang
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
Car detection is an important task that serves as a crucial prerequisite for many automated driving functions. The large variations in lighting/weather conditions and vehicle densities of the scenes pose significant challenges to existing car detection algorithms to meet the highly accurate perception demand for safety, due to the unstable/limited color information, which impedes the extraction of meaningful/discriminative features of cars. In this work, we present a novel learning-based car detection method that leverages trichromatic linear polarization as an additional cue to disambiguate such challenging cases. A key observation is that polarization, characteristic of the light wave, can robustly describe intrinsic physical properties of the scene objects in various imaging conditions and is strongly linked to the nature of materials for cars (e.g., metal and glass) and their surrounding environment (e.g., soil and trees), thereby providing reliable and discriminative features for robust car detection in challenging scenes. To exploit polarization cues, we first construct a pixel-aligned RGB-Polarization car detection dataset, which we subsequently employ to train a novel multimodal fusion network. Our car detection network dynamically integrates RGB and polarization features in a request-and-complement manner and can explore the intrinsic material properties of cars across all learning samples. We extensively validate our method and demonstrate that it outperforms state-of-the-art detection methods. Experimental results show that polarization is a powerful cue for car detection.
MOODv2: Masked Image Modeling for Out-of-Distribution Detection
Abstract
The crux of effective out-of-distribution (OOD) detection lies in acquiring a robust in-distribution (ID) representation, distinct from OOD samples. While previous methods predominantly leaned on recognition-based techniques for this purpose, they often resulted in shortcut learning, lacking comprehensive representations. In our study, we conducted a comprehensive analysis, exploring distinct pretraining tasks and employing various OOD score functions. The results highlight that the feature representations pre-trained through reconstruction yield a notable enhancement and narrow the performance gap among various score functions. This suggests that even simple score functions can rival complex ones when leveraging reconstruction-based pretext tasks. Reconstruction-based pretext tasks adapt well to various score functions. As such, it holds promising potential for further expansion. Our OOD detection framework, MOODv2, employs the masked image modeling pretext task. Without bells and whistles, MOODv2 impressively enhances 14.30% AUROC to 95.68% on ImageNet and achieves 99.98% on CIFAR-10.
Characteristics and prevalence of fake social media profiles with AI-generated faces
Abstract
Recent advancements in generative artificial intelligence (AI) have raised concerns about their potential to create convincing fake social media accounts, but empirical evidence is lacking. In this paper, we present a systematic analysis of Twitter(X) accounts using human faces generated by Generative Adversarial Networks (GANs) for their profile pictures. We present a dataset of 1,353 such accounts and show that they are used to spread scams, spam, and amplify coordinated messages, among other inauthentic activities. Leveraging a feature of GAN-generated faces -- consistent eye placement -- and supplementing it with human annotation, we devise an effective method for identifying GAN-generated profiles in the wild. Applying this method to a random sample of active Twitter users, we estimate a lower bound for the prevalence of profiles using GAN-generated faces between 0.021% and 0.044% -- around 10K daily active accounts. These findings underscore the emerging threats posed by multimodal generative AI. We release the source code of our detection method and the data we collect to facilitate further investigation. Additionally, we provide practical heuristics to assist social media users in recognizing such accounts.
Exception-aware Lifecycle Model Construction for Framework APIs
Authors: Authors: Jiwei Yan, Jinhao Huang, Hengqin Yang, Jun Yan
Abstract
The implementation of complex software systems usually depends on low-level frameworks or third-party libraries. During their evolution, the APIs adding and removing behaviors may cause unexpected compatibility problems. So, precisely analyzing and constructing the framework/ library's API lifecycle model is of great importance. Existing works have proposed the API existence-changing model for defect detection, while not considering the influence of semantic changes in APIs. In some cases, developers will not remove or deprecate APIs but modify their semantics by adding, removing, or modifying their exception-thrown code, which may bring potential defects to upper-level code. Therefore, besides the API existence model, it is also necessary for developers to be concerned with the exception-related code evolution in APIs, which requires the construction of exception-aware API lifecycle models for framework/library projects. To achieve automatic exception-aware API lifecycle model construction, this paper adopts a static analysis technique to extract exception summary information in the framework API code and adopts a multi-step matching strategy to obtain the changing process of exceptions. Then, it generates exception-aware API lifecycle models for the given framework/library project. With this approach, the API lifecycle extraction tool, JavaExP, is implemented, which is based on Java bytecode analysis. Compared to the state-of-the-art tool, JavaExP achieves both a higher F1 score (+60%) and efficiency (+7x), whose precision of exception matching and changing results is 98%. Compared to the exception-unaware API lifecycle modeling on 60 versions, JavaExp can identify 18% times more API changes. Among the 75,433 APIs under analysis, 20% of APIs have changed their exception-throwing behavior at least once after API introduction, which may bring many hidden compatibility issues.
Message Feedback Interference Cancellation Aided UAMP Iterative Detector for OTFS Systems
Abstract
The designing of efficient signal detectors is important and yet challenge for orthogonal time frequency space (OTFS) systems in high-mobility scenarios. In this letter, we develop an efficient message feedback interference cancellation aided unitary approximate message passing (denoted as UAMPMFIC) iterative detector, where the latest feedback messages from variable nodes are utilized for more reliable interference cancellation and performance improvement. A fast recursive scheme is leveraged in the proposed UAMP-MFIC detector to prevent complexity increasing. To further alleviate the error-propagation and improve the receiver performance, we also develop the bidirectional symbol detection structures, where Turbo UAMP-MFIC detector and iterative weight UAMP-MFIC detector are proposed to efficiently fuse the estimation results of forward and backward UAMP-MFIC detectors. The simulation results are finally provided to demonstrate performance improvement of our proposed detectors over existing detectors.
VoxelNextFusion: A Simple, Unified and Effective Voxel Fusion Framework for Multi-Modal 3D Object Detection
Authors: Authors: Ziying Song, Guoxin Zhang, Jun Xie, Lin Liu, Caiyan Jia, Shaoqing Xu, Zhepeng Wang
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
LiDAR-camera fusion can enhance the performance of 3D object detection by utilizing complementary information between depth-aware LiDAR points and semantically rich images. Existing voxel-based methods face significant challenges when fusing sparse voxel features with dense image features in a one-to-one manner, resulting in the loss of the advantages of images, including semantic and continuity information, leading to sub-optimal detection performance, especially at long distances. In this paper, we present VoxelNextFusion, a multi-modal 3D object detection framework specifically designed for voxel-based methods, which effectively bridges the gap between sparse point clouds and dense images. In particular, we propose a voxel-based image pipeline that involves projecting point clouds onto images to obtain both pixel- and patch-level features. These features are then fused using a self-attention to obtain a combined representation. Moreover, to address the issue of background features present in patches, we propose a feature importance module that effectively distinguishes between foreground and background features, thus minimizing the impact of the background features. Extensive experiments were conducted on the widely used KITTI and nuScenes 3D object detection benchmarks. Notably, our VoxelNextFusion achieved around +3.20% in AP@0.7 improvement for car detection in hard level compared to the Voxel R-CNN baseline on the KITTI test dataset
The Vulnerability Is in the Details: Locating Fine-grained Information of Vulnerable Code Identified by Graph-based Detectors
Authors: Authors: Baijun Cheng, Kailong Wang, Cuiyun Gao, Xiapu Luo, Yulei Sui, Li Li, Yao Guo, Xiangqun Chen, Haoyu Wang
Abstract
Vulnerability detection is a crucial component in the software development lifecycle. Existing vulnerability detectors, especially those based on deep learning (DL) models, have achieved high effectiveness. Despite their capability of detecting vulnerable code snippets from given code fragments, the detectors are typically unable to further locate the fine-grained information pertaining to the vulnerability, such as the precise vulnerability triggering locations.In this paper, we propose VULEXPLAINER, a tool for automatically locating vulnerability-critical code lines from coarse-level vulnerable code snippets reported by DL-based detectors.Our approach takes advantage of the code structure and the semantics of the vulnerabilities. Specifically, we leverage program slicing to get a set of critical program paths containing vulnerability-triggering and vulnerability-dependent statements and rank them to pinpoint the most important one (i.e., sub-graph) as the data flow associated with the vulnerability. We demonstrate that VULEXPLAINER performs consistently well on four state-of-the-art graph-representation(GP)-based vulnerability detectors, i.e., it can flag the vulnerability-triggering code statements with an accuracy of around 90% against eight common C/C++ vulnerabilities, outperforming five widely used GNN-based explanation approaches. The experimental results demonstrate the effectiveness of VULEXPLAINER, which provides insights into a promising research line: integrating program slicing and deep learning for the interpretation of vulnerable code fragments.
Reading Between the Frames: Multi-Modal Depression Detection in Videos from Non-Verbal Cues
Authors: Authors: David Gimeno-Gómez, Ana-Maria Bucur, Adrian Cosma, Carlos-David Martínez-Hinarejos, Paolo Rosso
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
Depression, a prominent contributor to global disability, affects a substantial portion of the population. Efforts to detect depression from social media texts have been prevalent, yet only a few works explored depression detection from user-generated video content. In this work, we address this research gap by proposing a simple and flexible multi-modal temporal model capable of discerning non-verbal depression cues from diverse modalities in noisy, real-world videos. We show that, for in-the-wild videos, using additional high-level non-verbal cues is crucial to achieving good performance, and we extracted and processed audio speech embeddings, face emotion embeddings, face, body and hand landmarks, and gaze and blinking information. Through extensive experiments, we show that our model achieves state-of-the-art results on three key benchmark datasets for depression detection from video by a substantial margin. Our code is publicly available on GitHub.
Systematic review of image segmentation using complex networks
Authors: Authors: Amin Rezaei, Fatemeh Asadi
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
This review presents various image segmentation methods using complex networks. Image segmentation is one of the important steps in image analysis as it helps analyze and understand complex images. At first, it has been tried to classify complex networks based on how it being used in image segmentation. In computer vision and image processing applications, image segmentation is essential for analyzing complex images with irregular shapes, textures, or overlapping boundaries. Advanced algorithms make use of machine learning, clustering, edge detection, and region-growing techniques. Graph theory principles combined with community detection-based methods allow for more precise analysis and interpretation of complex images. Hybrid approaches combine multiple techniques for comprehensive, robust segmentation, improving results in computer vision and image processing tasks.
Weakly Semi-supervised Tool Detection in Minimally Invasive Surgery Videos
Abstract
Surgical tool detection is essential for analyzing and evaluating minimally invasive surgery videos. Current approaches are mostly based on supervised methods that require large, fully instance-level labels (i.e., bounding boxes). However, large image datasets with instance-level labels are often limited because of the burden of annotation. Thus, surgical tool detection is important when providing image-level labels instead of instance-level labels since image-level annotations are considerably more time-efficient than instance-level annotations. In this work, we propose to strike a balance between the extremely costly annotation burden and detection performance. We further propose a co-occurrence loss, which considers a characteristic that some tool pairs often co-occur together in an image to leverage image-level labels. Encapsulating the knowledge of co-occurrence using the co-occurrence loss helps to overcome the difficulty in classification that originates from the fact that some tools have similar shapes and textures. Extensive experiments conducted on the Endovis2018 dataset in various data settings show the effectiveness of our method.
DocGraphLM: Documental Graph Language Model for Information Extraction
Authors: Authors: Dongsheng Wang, Zhiqiang Ma, Armineh Nourbakhsh, Kang Gu, Sameena Shah
Subjects: Computation and Language (cs.CL); Information Retrieval (cs.IR)
Abstract
Advances in Visually Rich Document Understanding (VrDU) have enabled information extraction and question answering over documents with complex layouts. Two tropes of architectures have emerged -- transformer-based models inspired by LLMs, and Graph Neural Networks. In this paper, we introduce DocGraphLM, a novel framework that combines pre-trained language models with graph semantics. To achieve this, we propose 1) a joint encoder architecture to represent documents, and 2) a novel link prediction approach to reconstruct document graphs. DocGraphLM predicts both directions and distances between nodes using a convergent joint loss function that prioritizes neighborhood restoration and downweighs distant node detection. Our experiments on three SotA datasets show consistent improvement on IE and QA tasks with the adoption of graph features. Moreover, we report that adopting the graph features accelerates convergence in the learning process during training, despite being solely constructed through link prediction.
Keyword: face recognition
There is no result
Keyword: augmentation
Image-based Deep Learning for Smart Digital Twins: a Review
Authors: Authors: Md Ruman Islam, Mahadevan Subramaniam, Pei-Chi Huang (Department of Computer Science, University of Nebraska at Omaha, Omaha, NE, USA)
Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Systems and Control (eess.SY)
Abstract
Smart Digital twins (SDTs) are being increasingly used to virtually replicate and predict the behaviors of complex physical systems through continual data assimilation enabling the optimization of the performance of these systems by controlling the actions of systems. Recently, deep learning (DL) models have significantly enhanced the capabilities of SDTs, particularly for tasks such as predictive maintenance, anomaly detection, and optimization. In many domains, including medicine, engineering, and education, SDTs use image data (image-based SDTs) to observe and learn system behaviors and control their behaviors. This paper focuses on various approaches and associated challenges in developing image-based SDTs by continually assimilating image data from physical systems. The paper also discusses the challenges involved in designing and implementing DL models for SDTs, including data acquisition, processing, and interpretation. In addition, insights into the future directions and opportunities for developing new image-based DL approaches to develop robust SDTs are provided. This includes the potential for using generative models for data augmentation, developing multi-modal DL models, and exploring the integration of DL with other technologies, including 5G, edge computing, and IoT. In this paper, we describe the image-based SDTs, which enable broader adoption of the digital twin DT paradigms across a broad spectrum of areas and the development of new methods to improve the abilities of SDTs in replicating, predicting, and optimizing the behavior of complex systems.
Unsupervised hard Negative Augmentation for contrastive learning
Abstract
We present Unsupervised hard Negative Augmentation (UNA), a method that generates synthetic negative instances based on the term frequency-inverse document frequency (TF-IDF) retrieval model. UNA uses TF-IDF scores to ascertain the perceived importance of terms in a sentence and then produces negative samples by replacing terms with respect to that. Our experiments demonstrate that models trained with UNA improve the overall performance in semantic textual similarity tasks. Additional performance gains are obtained when combining UNA with the paraphrasing augmentation. Further results show that our method is compatible with different backbone models. Ablation studies also support the choice of having a TF-IDF-driven control on negative augmentation.
Graph-level Protein Representation Learning by Structure Knowledge Refinement
Authors: Authors: Ge Wang, Zelin Zang, Jiangbin Zheng, Jun Xia, Stan Z. Li
Abstract
This paper focuses on learning representation on the whole graph level in an unsupervised manner. Learning graph-level representation plays an important role in a variety of real-world issues such as molecule property prediction, protein structure feature extraction, and social network analysis. The mainstream method is utilizing contrastive learning to facilitate graph feature extraction, known as Graph Contrastive Learning (GCL). GCL, although effective, suffers from some complications in contrastive learning, such as the effect of false negative pairs. Moreover, augmentation strategies in GCL are weakly adaptive to diverse graph datasets. Motivated by these problems, we propose a novel framework called Structure Knowledge Refinement (SKR) which uses data structure to determine the probability of whether a pair is positive or negative. Meanwhile, we propose an augmentation strategy that naturally preserves the semantic meaning of the original data and is compatible with our SKR framework. Furthermore, we illustrate the effectiveness of our SKR framework through intuition and experiments. The experimental results on the tasks of graph-level classification demonstrate that our SKR framework is superior to most state-of-the-art baselines.
Fus-MAE: A cross-attention-based data fusion approach for Masked Autoencoders in remote sensing
Authors: Authors: Hugo Chan-To-Hing, Bharadwaj Veeravalli
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
Self-supervised frameworks for representation learning have recently stirred up interest among the remote sensing community, given their potential to mitigate the high labeling costs associated with curating large satellite image datasets. In the realm of multimodal data fusion, while the often used contrastive learning methods can help bridging the domain gap between different sensor types, they rely on data augmentations techniques that require expertise and careful design, especially for multispectral remote sensing data. A possible but rather scarcely studied way to circumvent these limitations is to use a masked image modelling based pretraining strategy. In this paper, we introduce Fus-MAE, a self-supervised learning framework based on masked autoencoders that uses cross-attention to perform early and feature-level data fusion between synthetic aperture radar and multispectral optical data - two modalities with a significant domain gap. Our empirical findings demonstrate that Fus-MAE can effectively compete with contrastive learning strategies tailored for SAR-optical data fusion and outperforms other masked-autoencoders frameworks trained on a larger corpus.
From LLM to Conversational Agent: A Memory Enhanced Architecture with Fine-Tuning of Large Language Models
Authors: Authors: Na Liu, Liangyu Chen, Xiaoyu Tian, Wei Zou, Kaijiang Chen, Ming Cui
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
Abstract
This paper introduces RAISE (Reasoning and Acting through Scratchpad and Examples), an advanced architecture enhancing the integration of Large Language Models (LLMs) like GPT-4 into conversational agents. RAISE, an enhancement of the ReAct framework, incorporates a dual-component memory system, mirroring human short-term and long-term memory, to maintain context and continuity in conversations. It entails a comprehensive agent construction scenario, including phases like Conversation Selection, Scene Extraction, CoT Completion, and Scene Augmentation, leading to the LLMs Training phase. This approach appears to enhance agent controllability and adaptability in complex, multi-turn dialogues. Our preliminary evaluations in a real estate sales context suggest that RAISE has some advantages over traditional agents, indicating its potential for broader applications. This work contributes to the AI field by providing a robust framework for developing more context-aware and versatile conversational agents.
Object-Centric Instruction Augmentation for Robotic Manipulation
Abstract
Humans interpret scenes by recognizing both the identities and positions of objects in their observations. For a robot to perform tasks such as \enquote{pick and place}, understanding both what the objects are and where they are located is crucial. While the former has been extensively discussed in the literature that uses the large language model to enrich the text descriptions, the latter remains underexplored. In this work, we introduce the \textit{Object-Centric Instruction Augmentation (OCI)} framework to augment highly semantic and information-dense language instruction with position cues. We utilize a Multi-modal Large Language Model (MLLM) to weave knowledge of object locations into natural language instruction, thus aiding the policy network in mastering actions for versatile manipulation. Additionally, we present a feature reuse mechanism to integrate the vision-language features from off-the-shelf pre-trained MLLM into policy networks. Through a series of simulated and real-world robotic tasks, we demonstrate that robotic manipulator imitation policies trained with our enhanced instructions outperform those relying solely on traditional language instructions.
Plug-in Diffusion Model for Sequential Recommendation
Authors: Authors: Haokai Ma, Ruobing Xie, Lei Meng, Xin Chen, Xu Zhang, Leyu Lin, Zhanhui Kang
Abstract
Pioneering efforts have verified the effectiveness of the diffusion models in exploring the informative uncertainty for recommendation. Considering the difference between recommendation and image synthesis tasks, existing methods have undertaken tailored refinements to the diffusion and reverse process. However, these approaches typically use the highest-score item in corpus for user interest prediction, leading to the ignorance of the user's generalized preference contained within other items, thereby remaining constrained by the data sparsity issue. To address this issue, this paper presents a novel Plug-in Diffusion Model for Recommendation (PDRec) framework, which employs the diffusion model as a flexible plugin to jointly take full advantage of the diffusion-generating user preferences on all items. Specifically, PDRec first infers the users' dynamic preferences on all items via a time-interval diffusion model and proposes a Historical Behavior Reweighting (HBR) mechanism to identify the high-quality behaviors and suppress noisy behaviors. In addition to the observed items, PDRec proposes a Diffusion-based Positive Augmentation (DPA) strategy to leverage the top-ranked unobserved items as the potential positive samples, bringing in informative and diverse soft signals to alleviate data sparsity. To alleviate the false negative sampling issue, PDRec employs Noise-free Negative Sampling (NNS) to select stable negative samples for ensuring effective model optimization. Extensive experiments and analyses on four datasets have verified the superiority of the proposed PDRec over the state-of-the-art baselines and showcased the universality of PDRec as a flexible plugin for commonly-used sequential encoders in different recommendation scenarios. The code is available in https://github.com/hulkima/PDRec.
Keyword: detection
Locally Differentially Private Embedding Models in Distributed Fraud Prevention Systems
A comprehensive survey of research towards AI-enabled unmanned aerial systems in pre-, active-, and post-wildfire management
Image-based Deep Learning for Smart Digital Twins: a Review
Autonomous Multi-Rotor UAVs: A Holistic Approach to Design, Optimization, and Fabrication
A Community Detection and Graph Neural Network Based Link Prediction Approach for Scientific Literature
Exploiting Polarized Material Cues for Robust Car Detection
MOODv2: Masked Image Modeling for Out-of-Distribution Detection
Characteristics and prevalence of fake social media profiles with AI-generated faces
Exception-aware Lifecycle Model Construction for Framework APIs
Message Feedback Interference Cancellation Aided UAMP Iterative Detector for OTFS Systems
VoxelNextFusion: A Simple, Unified and Effective Voxel Fusion Framework for Multi-Modal 3D Object Detection
The Vulnerability Is in the Details: Locating Fine-grained Information of Vulnerable Code Identified by Graph-based Detectors
Reading Between the Frames: Multi-Modal Depression Detection in Videos from Non-Verbal Cues
Systematic review of image segmentation using complex networks
Weakly Semi-supervised Tool Detection in Minimally Invasive Surgery Videos
DocGraphLM: Documental Graph Language Model for Information Extraction
Keyword: face recognition
There is no result
Keyword: augmentation
Image-based Deep Learning for Smart Digital Twins: a Review
Unsupervised hard Negative Augmentation for contrastive learning
Graph-level Protein Representation Learning by Structure Knowledge Refinement
Fus-MAE: A cross-attention-based data fusion approach for Masked Autoencoders in remote sensing
From LLM to Conversational Agent: A Memory Enhanced Architecture with Fine-Tuning of Large Language Models
Object-Centric Instruction Augmentation for Robotic Manipulation
Plug-in Diffusion Model for Sequential Recommendation