DEtection TRansformer (DETR) and its variants (DETRs) have been successfully applied to crowded pedestrian detection, which achieved promising performance. However, we find that, in different degrees of crowded scenes, the number of DETRs' queries must be adjusted manually, otherwise, the performance would degrade to varying degrees. In this paper, we first analyze the two current query generation methods and summarize four guidelines for designing the adaptive query generation method. Then, we propose Rank-based Adaptive Query Generation (RAQG) to alleviate the problem. Specifically, we design a rank prediction head that can predict the rank of the lowest confidence positive training sample produced by the encoder. Based on the predicted rank, we design an adaptive selection method that can adaptively select coarse detection results produced by the encoder to generate queries. Moreover, to train the rank prediction head better, we propose Soft Gradient L1 Loss. The gradient of Soft Gradient L1 Loss is continuous, which can describe the relationship between the loss value and the updated value of model parameters granularly. Our method is simple and effective, which can be plugged into any DETRs to make it query-adaptive in theory. The experimental results on Crowdhuman dataset and Citypersons dataset show that our method can adaptively generate queries for DETRs and achieve competitive results. Especially, our method achieves state-of-the-art 39.4% MR on Crowdhuman dataset.
相關內容
Graph Neural Networks (GNNs) have emerged as a notorious alternative to address learning problems dealing with non-Euclidean datasets. However, although most works assume that the graph is perfectly known, the observed topology is prone to errors stemming from observational noise, graph-learning limitations, or adversarial attacks. If ignored, these perturbations may drastically hinder the performance of GNNs. To address this limitation, this work proposes a robust implementation of GNNs that explicitly accounts for the presence of perturbations in the observed topology. For any task involving GNNs, our core idea is to i) solve an optimization problem not only over the learnable parameters of the GNN but also over the true graph, and ii) augment the fitting cost with a term accounting for discrepancies on the graph. Specifically, we consider a convolutional GNN based on graph filters and follow an alternating optimization approach to handle the (non-differentiable and constrained) optimization problem by combining gradient descent and projected proximal updates. The resulting algorithm is not limited to a particular type of graph and is amenable to incorporating prior information about the perturbations. Finally, we assess the performance of the proposed method through several numerical experiments.
The growing presence of Artificial Intelligence (AI) in various sectors necessitates systems that accurately reflect societal diversity. This study seeks to envision the operationalization of the ethical imperatives of diversity and inclusion (D&I) within AI ecosystems, addressing the current disconnect between ethical guidelines and their practical implementation. A significant challenge in AI development is the effective operationalization of D&I principles, which is critical to prevent the reinforcement of existing biases and ensure equity across AI applications. This paper proposes a vision of a framework for developing a tool utilizing persona-based simulation by Generative AI (GenAI). The approach aims to facilitate the representation of the needs of diverse users in the requirements analysis process for AI software. The proposed framework is expected to lead to a comprehensive persona repository with diverse attributes that inform the development process with detailed user narratives. This research contributes to the development of an inclusive AI paradigm that ensures future technological advances are designed with a commitment to the diverse fabric of humanity.
Wave propagation problems are typically formulated as partial differential equations (PDEs) on unbounded domains to be solved. The classical approach to solving such problems involves truncating them to problems on bounded domains by designing the artificial boundary conditions or perfectly matched layers, which typically require significant effort, and the presence of nonlinearity in the equation makes such designs even more challenging. Emerging deep learning-based methods for solving PDEs, with the physics-informed neural networks (PINNs) method as a representative, still face significant challenges when directly used to solve PDEs on unbounded domains. Calculations performed in a bounded domain of interest without imposing boundary constraints can lead to a lack of unique solutions thus causing the failure of PINNs. In light of this, this paper proposes a novel and effective operator learning-based method for solving PDEs on unbounded domains. The key idea behind this method is to generate high-quality training data. Specifically, we construct a family of approximate analytical solutions to the target PDE based on its initial condition and source term. Then, using these constructed data comprising exact solutions, initial conditions, and source terms, we train an operator learning model called MIONet, which is capable of handling multiple inputs, to learn the mapping from the initial condition and source term to the PDE solution on a bounded domain of interest. Finally, we utilize the generalization ability of this model to predict the solution of the target PDE. The effectiveness of this method is exemplified by solving the wave equation and the Schrodinger equation defined on unbounded domains. More importantly, the proposed method can deal with nonlinear problems, which has been demonstrated by solving Burger's equation and Korteweg-de Vries (KdV) equation.
Sarcasm is a form of irony that requires readers or listeners to interpret its intended meaning by considering context and social cues. Machine learning classification models have long had difficulty detecting sarcasm due to its social complexity and contradictory nature. This paper explores the applications of the Generative Pretrained Transformer (GPT) models, including GPT-3, InstructGPT, GPT-3.5, and GPT-4, in detecting sarcasm in natural language. It tests fine-tuned and zero-shot models of different sizes and releases. The GPT models were tested on the political and balanced (pol-bal) portion of the popular Self-Annotated Reddit Corpus (SARC 2.0) sarcasm dataset. In the fine-tuning case, the largest fine-tuned GPT-3 model achieves accuracy and $F_1$-score of 0.81, outperforming prior models. In the zero-shot case, one of GPT-4 models yields an accuracy of 0.70 and $F_1$-score of 0.75. Other models score lower. Additionally, a model's performance may improve or deteriorate with each release, highlighting the need to reassess performance after each release.
We developed a statistical theory of zero-count-detector (ZCD), which is defined as a zero-class Poisson under conditions outlined in the paper. ZCD is often encountered in the studies of rare events in physics, health physics, and many other fields where counting of events occurs. We found no acceptable solution to ZCD in classical statistics and affirmed the need for the Bayesian statistics. Several uniform and reference priors were studied and we derived Bayesian posteriors, point estimates, and upper limits. It was showed that the maximum-entropy prior, containing the most information, resulted in the smallest bias and the lowest risk, making it the most admissible and acceptable among the priors studied. We also investigated application of zero-inflated Poisson and Negative-binomial distributions to ZCD. It was showed using Bayesian marginalization that, under limited information, these distributions reduce to the Poisson distribution.
Conventional entity typing approaches are based on independent classification paradigms, which make them difficult to recognize inter-dependent, long-tailed and fine-grained entity types. In this paper, we argue that the implicitly entailed extrinsic and intrinsic dependencies between labels can provide critical knowledge to tackle the above challenges. To this end, we propose \emph{Label Reasoning Network(LRN)}, which sequentially reasons fine-grained entity labels by discovering and exploiting label dependencies knowledge entailed in the data. Specifically, LRN utilizes an auto-regressive network to conduct deductive reasoning and a bipartite attribute graph to conduct inductive reasoning between labels, which can effectively model, learn and reason complex label dependencies in a sequence-to-set, end-to-end manner. Experiments show that LRN achieves the state-of-the-art performance on standard ultra fine-grained entity typing benchmarks, and can also resolve the long tail label problem effectively.
Graph Convolution Networks (GCNs) manifest great potential in recommendation. This is attributed to their capability on learning good user and item embeddings by exploiting the collaborative signals from the high-order neighbors. Like other GCN models, the GCN based recommendation models also suffer from the notorious over-smoothing problem - when stacking more layers, node embeddings become more similar and eventually indistinguishable, resulted in performance degradation. The recently proposed LightGCN and LR-GCN alleviate this problem to some extent, however, we argue that they overlook an important factor for the over-smoothing problem in recommendation, that is, high-order neighboring users with no common interests of a user can be also involved in the user's embedding learning in the graph convolution operation. As a result, the multi-layer graph convolution will make users with dissimilar interests have similar embeddings. In this paper, we propose a novel Interest-aware Message-Passing GCN (IMP-GCN) recommendation model, which performs high-order graph convolution inside subgraphs. The subgraph consists of users with similar interests and their interacted items. To form the subgraphs, we design an unsupervised subgraph generation module, which can effectively identify users with common interests by exploiting both user feature and graph structure. To this end, our model can avoid propagating negative information from high-order neighbors into embedding learning. Experimental results on three large-scale benchmark datasets show that our model can gain performance improvement by stacking more layers and outperform the state-of-the-art GCN-based recommendation models significantly.
Graph Convolutional Network (GCN) has been widely applied in transportation demand prediction due to its excellent ability to capture non-Euclidean spatial dependence among station-level or regional transportation demands. However, in most of the existing research, the graph convolution was implemented on a heuristically generated adjacency matrix, which could neither reflect the real spatial relationships of stations accurately, nor capture the multi-level spatial dependence of demands adaptively. To cope with the above problems, this paper provides a novel graph convolutional network for transportation demand prediction. Firstly, a novel graph convolution architecture is proposed, which has different adjacency matrices in different layers and all the adjacency matrices are self-learned during the training process. Secondly, a layer-wise coupling mechanism is provided, which associates the upper-level adjacency matrix with the lower-level one. It also reduces the scale of parameters in our model. Lastly, a unitary network is constructed to give the final prediction result by integrating the hidden spatial states with gated recurrent unit, which could capture the multi-level spatial dependence and temporal dynamics simultaneously. Experiments have been conducted on two real-world datasets, NYC Citi Bike and NYC Taxi, and the results demonstrate the superiority of our model over the state-of-the-art ones.
Graph Neural Networks (GNNs) have recently become increasingly popular due to their ability to learn complex systems of relations or interactions arising in a broad spectrum of problems ranging from biology and particle physics to social networks and recommendation systems. Despite the plethora of different models for deep learning on graphs, few approaches have been proposed thus far for dealing with graphs that present some sort of dynamic nature (e.g. evolving features or connectivity over time). In this paper, we present Temporal Graph Networks (TGNs), a generic, efficient framework for deep learning on dynamic graphs represented as sequences of timed events. Thanks to a novel combination of memory modules and graph-based operators, TGNs are able to significantly outperform previous approaches being at the same time more computationally efficient. We furthermore show that several previous models for learning on dynamic graphs can be cast as specific instances of our framework. We perform a detailed ablation study of different components of our framework and devise the best configuration that achieves state-of-the-art performance on several transductive and inductive prediction tasks for dynamic graphs.
Attention mechanism has been used as an ancillary means to help RNN or CNN. However, the Transformer (Vaswani et al., 2017) recently recorded the state-of-the-art performance in machine translation with a dramatic reduction in training time by solely using attention. Motivated by the Transformer, Directional Self Attention Network (Shen et al., 2017), a fully attention-based sentence encoder, was proposed. It showed good performance with various data by using forward and backward directional information in a sentence. But in their study, not considered at all was the distance between words, an important feature when learning the local dependency to help understand the context of input text. We propose Distance-based Self-Attention Network, which considers the word distance by using a simple distance mask in order to model the local dependency without losing the ability of modeling global dependency which attention has inherent. Our model shows good performance with NLI data, and it records the new state-of-the-art result with SNLI data. Additionally, we show that our model has a strength in long sentences or documents.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191