Few-shot learning (FSL) poses a considerable challenge since it aims to improve the model generalization ability with limited labeled data. Previous works usually attempt to construct class-specific prototypes and then predict novel classes using these prototypes. However, the feature distribution represented by the limited labeled data is coarse-grained, leading to large information gap between the labeled and unlabeled data as well as biases in the prototypes. In this paper, we investigate the correlation between sample quality and density, and propose a Density-driven Iterative Prototype Optimization to acquire high-quality prototypes, and further improve few-shot learning performance. Specifically, the proposed method consists of two optimization strategies. The similarity-evaluating strategy is for capturing the information gap between the labeled and unlabeled data by reshaping the feature manifold for the novel feature distribution. The density-driven strategy is proposed to iteratively refine the prototypes in the direction of density growth. The proposed method could reach or even exceed the state-of-the-art performance on four benchmark datasets, including mini-ImageNet, tiered-ImageNet, CUB, and CIFAR-FS. The code will be available soon at https://github.com/tailofcat/DIPO.
论文检索
输入标题、作者或关键词，从 1,048 篇学术成果中精准定位
Neural trees benefit from the high-level representation of neural networks and the interpretability of decision trees. Therefore, the existing works on neural trees perform outstandingly on various tasks such as architecture search. However, these works require every router to provide only one successor for each sample, causing the predictions to be dominated by the elite branch and its derivative architectures. To break this branch dominance, we propose the combinatorial routing neural tree method, termed CombRo. Unlike the previous methods employing unicast routing, CombRo performs multicast schema in each iteration, allowing the features to be routed to any combination of successors at every non-leaf. The weights of each architecture are then evaluated accordingly. We update the weights by training the routing subnetwork, and the architecture with the top weight is selected in the final step. We compare CombRo with the existing algorithms on 3 public image datasets, demonstrating its superior performance in terms of accuracy. Visualization results further validate the effectiveness of the multicast routing schema. Code is available at https://github.com/JiahaoLi-gdut/CombRo.
Self-supervised learning (SSL) provides a promising alternative for representation learning on hypergraphs without costly labels. However, existing hypergraph SSL models are mostly based on contrastive methods with the instance-level discrimination strategy, suffering from two significant limitations: (1) They select negative samples arbitrarily, which is unreliable in deciding similar and dissimilar pairs, causing training bias. (2) They often require a large number of negative samples, resulting in expensive computational costs. To address the above issues, we propose SE-HSSL, a hypergraph SSL framework with three sampling-efficient self-supervised signals. Specifically, we introduce two sampling-free objectives leveraging the canonical correlation analysis as the node-level and group-level self-supervised signals. Additionally, we develop a novel hierarchical membership-level contrast objective motivated by the cascading overlap relationship in hypergraphs, which can further reduce membership sampling bias and improve the efficiency of sample utilization. Through comprehensive experiments on 7 real-world hypergraphs, we demonstrate the superiority of our approach over the state-of-the-art method in terms of both effectiveness and efficiency.
Active learning is known to be a well-motivated algorithm that aims to maximize model performance with relatively small data, but it introduces sampling bias due to active selection. To adjust the bias, current literature utilizes corrective weights in a supervised learning approach. However, those methods consider only a small amount of actively sampled data and thus estimation efficiency can be improved using unsampled data together. In this paper, we develop an actively improved augmented estimation equation (AI-AEE) based on corrective weights as well as imputation models that allow us to leverage unlabeled data. The asymptotic distribution of the proposed estimator as the solution to the AI-AEE is derived, and an optimal sampling scheme to minimize the asymptotic mean squared error of the estimator is proposed. We then propose a general practical algorithm for training prediction models in the active and semi-supervised learning framework. The superiority of our method is demonstrated on synthetic and real data examples.
Hyperbolic geometry provides an ideal setting to represent the scale-free or hierarchical characteristics of an input graph naturally. Utilizing hyperbolic geometry for learning dynamic graph representation has gained a growing interest in recent years. However, the majority of hyperbolic-based approaches rely on tangent spaces to perform graph operations, which could distort the structure of the dynamic graph when the graph grows over time. To avoid the distortion in tangent space, we propose a Hyperbolic Manifold-Preserving Temporal Graph Network that works directly on the hyperbolic manifold. The model includes a graph convolution module for learning the spatial dependencies, an attention architecture for capturing the temporal properties, and a gated recurrent unit for extracting the spatio-temporal relationships. By evaluating on diverse real-world dynamic graphs, our model has achieved significant improvements in link prediction and new link prediction tasks, in comparison with other baselines.
The extended Weber location problem is a classical optimization problem that has inspired some new works in several machine learning scenarios recently. However, most existing algorithms may get stuck due to the singularity at the data points when the power of the cost function 1\<= q<2, such as the widely-used iterative Weiszfeld approach. In this paper, we establish a de-singularity subgradient approach for this problem. We also provide a complete proof of convergence which has fixed some incomplete statements of the proofs for some previous Weiszfeld algorithms. Moreover, we deduce a new theoretical result of superlinear convergence for the iteration sequence in a special case where the minimum point is a singular point. We conduct extensive experiments in a real-world machine learning scenario to show that the proposed approach solves the singularity problem, produces the same results as in the non-singularity cases, and shows a reasonable rate of linear convergence. The results also indicate that the q-th power case (1<q<2) is more advantageous than the 1-st power case and the 2-nd power case in some situations. Hence the de-singularity subgradient approach is beneficial to advancing both theory and practice for the extended Weber location problem.
Deep Neural Networks (DNNs) can be represented as graphs whose links and vertices iteratively process data and solve tasks sub-optimally. Complex Network Theory (CNT), merging statistical physics with graph theory, provides a method for interpreting neural networks by analysing their weights and neuron structures. However, classic works adapt CNT metrics that only permit a topological analysis as they do not account for the effect of the input data. In addition, CNT metrics have been applied to a limited range of architectures, mainly including Fully Connected neural networks. In this work, we extend the existing CNT metrics with measures that sample from the DNNs' training distribution, shifting from a purely topological analysis to one that connects with the interpretability of deep learning. For the novel metrics, in addition to the existing ones, we provide a mathematical formalisation for Fully Connected, AutoEncoder, Convolutional and Recurrent neural networks, of which we vary the activation functions and the number of hidden layers. We show that these metrics differentiate DNNs based on the architecture, the number of hidden layers, and the activation function. Our contribution provides a method rooted in physics for interpreting DNNs that offers insights beyond the traditional input-output relationship and the CNT topological analysis.
Designing algorithms is cumbersome and error-prone. This, among other things, has increasingly led to efforts to extend or even replace designing algorithms with machine learning models. While previous research has demonstrated that some machine learning models possess Turing-completeness, the findings are largely theoretical, and solutions for specific algorithmic tasks remain unclear. With this in mind, we investigate the feasibility of learning representations of classical algorithms from data on their execution, enabling their application to different inputs. We propose a novel and general framework for algorithm learning consisting of a model of computation that facilitates algorithm analysis across various levels of abstraction. We formalize the problem of learning an algorithm using an algebraic approach for graph traversal. We apply this framework to comparison sorts and evaluate the inferred machine learning models' performance, demonstrating the applicability of the approach in terms of accuracy and sensitivity.
Search in test time is often used to improve the performance of reinforcement learning algorithms. Performing theoretically sound search in fully adversarial two-player games with imperfect information is notoriously difficult and requires a complicated training process. We present a method for adding test-time search to an arbitrary policy-gradient algorithm that learns from sampled trajectories. Besides the policy network, the algorithm trains an additional critic network, which estimates the expected values of players following various transformations of the policies given by the policy network. These values are then used for depth-limited search. We show how the values from this critic can create a value function for imperfect information games. Moreover, they can be used to compute the summary statistics necessary to start the search from an arbitrary decision point in the game. The presented algorithm is scalable to very large games since it does not require any search during train time. We evaluate the algorithm's performance when trained along Regularized Nash Dynamics, and we evaluate the benefit of using the search in the standard benchmark game of Leduc hold'em, multiple variants of imperfect information Goofspiel, and Battleships.
We consider the problem of class expression learning using cardinality-minimal sets of examples. Recent class expression learning approaches employ deep neural networks and have demonstrated tremendous performance improvements in execution time and quality of the computed solutions. However, they lack generalization capabilities when it comes to the number of examples used in a learning problem, i.e., they often perform poorly on unseen learning problems where only a few examples are given. In this work, we propose a generalization of the classical class expression learning problem to address the limitations above. In short, our generalized learning problem (GLP) forces learning systems to solve the classical class expression learning problem using the smallest possible subsets of examples, thereby improving the learning systems' ability to solve unseen learning problems with arbitrary numbers of examples. Moreover, we develop ROCES, a learning algorithm for synthesis-based approaches to solve GLP. Experimental results suggest that post training, ROCES outperforms existing synthesis-based approaches on out-of-distribution learning problems while remaining highly competitive overall.
Label Distribution Learning (LDL) is a novel machine learning paradigm that assigns label distribution to each instance. Numerous LDL methods proposed to leverage label correlation in the learning process to solve the exponential-sized output space; among these, many exploited the low-rank structure of label distribution to capture label correlation. However, recent research has unveiled that label distribution matrices typically maintain full rank, posing a challenge to approaches relying on low-rank label correlation. Notably, low-rank label correlation finds widespread adoption in multi-label learning (MLL) literature due to the often low-rank nature of multi-label matrices. Inspired by that, we introduce an auxiliary MLL process within the LDL framework, focusing on capturing low-rank label correlation within this auxiliary MLL component rather than the LDL itself. By doing so, we adeptly exploited low-rank label correlation in our LDL methods. We conduct comprehensive experiments and demonstrate that our methods are superior to existing LDL methods. Besides, the ablation studies justify the advantages of exploiting low-rank label correlation in the auxiliary MLL.
Efficient and Stable Offline-to-online Reinforcement Learning via Continual Policy Revitalization
PDF ↗In offline Reinforcement Learning (RL), the pre-trained policies are utilized for initialization and subsequent online fine-tuning. However, existing methods suffer from instability and low sample efficiency compared to pure online learning. This paper identifies these limitations stemming from direct policy initialization using offline-trained policy models. We propose Continual Policy Revitalization (CPR) as a novel efficient, stable fine-tuning method. CPR incorporates a periodic policy revitalization technique, restoring the overtrained policy network to full learning capacity while ensuring stable initial performance. This approach enables fine-tuning without being adversely affected by low-quality pre-trained policies. In contrast to previous research, CPR initializes the new policy with an adaptive policy constraint in policy optimization. Such optimization keeps the new policy close to behavior policy constructed from historical policies. This contributes to stable policy improvement and optimal converged performance. Practically, CPR can seamlessly integrate into existing offline RL algorithms with minimal modification. We empirically validate the effectiveness of our method through extensive experiments, demonstrating substantial improvements in learning stability and efficiency compared to previous approaches. Our code is available at https://github.com/LAMDA-RL/CPR.
Learning Causally Disentangled Representations via the Principle of Independent Causal Mechanisms
PDF ↗Learning disentangled causal representations is a challenging problem that has gained significant attention recently due to its implications for extracting meaningful information for downstream tasks. In this work, we define a new notion of causal disentanglement from the perspective of independent causal mechanisms. We propose ICM-VAE, a framework for learning causally disentangled representations supervised by causally related observed labels. We model causal mechanisms using nonlinear learnable flow-based diffeomorphic functions to map noise variables to latent causal variables. Further, to promote the disentanglement of causal factors, we propose a causal disentanglement prior learned from auxiliary labels and the latent causal structure. We theoretically show the identifiability of causal factors and mechanisms up to permutation and elementwise reparameterization. We empirically demonstrate that our framework induces highly disentangled causal factors, improves interventional robustness, and is compatible with counterfactual generation.
Data-driven offline reinforcement learning and imitation learning approaches have been gaining popularity in addressing sequential decision-making problems. Yet, these approaches rarely consider learning Pareto-optimal policies from a limited pool of expert datasets. This becomes particularly marked due to practical limitations in obtaining comprehensive datasets for all preferences, where multiple conflicting objectives exist and each expert might hold a unique optimization preference for these objectives. In this paper, we adapt inverse reinforcement learning (IRL) by using reward distance estimates for regularizing the discriminator. This enables progressive generation of a set of policies that accommodate diverse preferences on the multiple objectives, while using only two distinct datasets, each associated with a different expert preference. In doing so, we present a Pareto IRL framework (ParIRL) that establishes a Pareto policy set from these limited datasets. In the framework, the Pareto policy set is then distilled into a single, preference-conditioned diffusion model, thus allowing users to immediately specify which expert's patterns they prefer. Through experiments, we show that ParIRL outperforms other IRL algorithms for various multi-objective control tasks, achieving the dense approximation of the Pareto frontier. We also demonstrate the applicability of ParIRL with autonomous driving in CARLA.
Hybrid models that combine CNNs and ViTs have recently emerged as state-of-the-art computer vision models. To efficiently deploy these hybrid models on resource-constrained mobile/edge devices, quantization is emerging as a promising solution. However, post-training quantization (PTQ), which does not require retraining or labeled data, has not been extensively studied for hybrid models. In this study, we propose a novel PTQ technique specialized for CNN-transformer hybrid models by considering the hardware design of hybrid models on AI accelerators such as GPUs and FPGAs. First, we introduce quantization-aware distribution scaling to address the large outliers caused by inter-channel variance in convolution layers. Furthermore, in the transformer block, we propose approximating the integer-only softmax with a linear function. This approach allows us to avoid costly FP32/INT32 multiplications, resulting in more efficient computations. Experimental results show that the proposed quantization method with INT8 precision demonstrated a 0.39% accuracy drop compared with the FP32 baseline on MobileViT-s with the ImageNet-1k dataset. Furthermore, when implemented on the FPGA platform, the proposed linear softmax achieved significant resource savings, reducing the look-up table and flip-flop usage by 1.8 ~ 2.1x and 1.3 ~ 1.9x, respectively, compared with the existing second-order polynomial approximation. The code is available at https://github.com/IDSL-SeoulTech/HyQ.
Goal-conditioned (GC) policy learning often faces a challenge arising from the sparsity of rewards, when confronting long-horizon goals. To address the challenge, we explore skill-based GC policy learning in offline settings, where skills are acquired from existing data and long-horizon goals are decomposed into sequences of near-term goals that align with these skills. Specifically, we present an `offline GC policy learning via skill-step abstraction' framework (GLvSA) tailored for tackling long-horizon GC tasks affected by goal distribution shifts. In the framework, a GC policy is progressively learned offline in conjunction with the incremental modeling of skill-step abstractions on the data. We also devise a GC policy hierarchy that not only accelerates GC policy learning within the framework but also allows for parameter-efficient fine-tuning of the policy. Through experiments with the maze and Franka kitchen environments, we demonstrate the superiority and efficiency of our GLvSA framework in adapting GC policies to a wide range of long-horizon goals. The framework achieves competitive zero-shot and few-shot adaptation performance, outperforming existing GC policy learning and skill-based methods.
In recent years, a variety of graph learning-based multi-view clustering (MVC) methods have emerged. However, these methods continue to face challenges in extracting latent features from real-world data, particularly in scenarios involving high-resolution color images and high-dimensional features. This task is notably difficult in cases where images are visually similar yet semantically diverse. To address this issue, we present a novel large-scale pre-trained model for multi-view clustering, named Integrate Vision-Language Semantic Graphs in Multi-View Clustering (IVSGMV), which harnesses the capabilities of visual-language pre-training models to enhance clustering performance and confronts issues in the unsupervised tuning of pre-trained models for multi-view data. We introduce an effective unsupervised approach for creating semantic graphs from image multi-view datasets using pre-trained encoders. Our method addresses the inherent spatial noise and imbalance in these encoders by employing graph filters and a joint process that integrates both image node and edge features. Additionally, we demonstrate the application of our approach to multi-view image clustering on extensive datasets, notably the high-resolution MVImgNet, achieving an impressive 82% accuracy. Furthermore, our method extends the zero-shot capabilities of large-scale pre-trained models, resulting in good performance in clustering tasks on untrained multi-view datasets.
In a regression task, a function is learned from labeled data to predict the labels at new data points. The goal is to achieve small prediction errors. In symbolic regression, the goal is more ambitious, namely, to learn an interpretable function that makes small prediction errors. This additional goal largely rules out the standard approach used in regression, that is, reducing the learning problem to learning parameters of an expansion of basis functions by optimization. Instead, symbolic regression methods search for a good solution in a space of symbolic expressions. To cope with the typically vast search space, most symbolic regression methods make implicit, or sometimes even explicit, assumptions about its structure. Here, we argue that the only obvious structure of the search space is that it contains small expressions, that is, expressions that can be decomposed into a few subexpressions. We show that systematically searching spaces of small expressions finds solutions that are more accurate and more robust against noise than those obtained by state-of-the-art symbolic regression methods. In particular, systematic search outperforms state-of-the-art symbolic regressors in terms of its ability to recover the true underlying symbolic expressions on established benchmark data sets.
Temporal Domain Generalization (TDG) aims at learning models under temporally evolving data distributions and achieving generalization to unseen future data distributions following the evolving trend. Existing advanced TDG methods learn the evolving patterns through the collective behaviors observed at the population-level of instances, such as time-varying statistics and parameters, tending to overlook the impact of individual-level instance evolving processes on the decision boundary. However, a major obstacle is that datasets at different timestamps may comprise unrelated instances and there is no inherent existence of the instance-level evolving trajectories, which hinders us from learning how the decision boundary changes. To address the above challenges, we propose a Continuous-Time modelling Optimal Transport trajectories (CTOT) framework in this paper. Specifically, we utilize optimal transport to align the data distributions between each pair of adjacent source domains to construct instance evolving trajectories. Subsequently, they are modelled by a continuous-time model and extrapolated to generate future virtual instances, which help the model to adapt its decision boundary to the future domain. Extensive experiments on multiple classification and regression benchmarks demonstrate the effectiveness of the proposed CTOT framework. The code and appendix are both available on https://github.com/JinYujie99/CTOT.
Because of the wide deployment of smart contracts, smart contract vulnerabilities pose a challenging risk to blockchain security. Currently, deep learning-based vulnerability detection is a very attractive solution due to its ability to identify complex patterns and features. The existing methods mainly consider the contract code content features, expert knowledge patterns, and contract code modalities. To further enhance smart contract vulnerability detection, this paper attempts to identify community features from smart contracts with similar semantic and syntactic structures, and shared features from two related vulnerability detection tasks, vulnerability classification and localization. The experimental results verify that the proposed approach significantly outperforms the state-of-the-art methods in terms of accuracy, recall, precision, and F1-score.