Vision-language models (VLMs) have emerged as formidable tools, showing their strong capability in handling various open-vocabulary tasks in image recognition, text-driven visual content generation, and visual chatbots, to name a few. In recent years, considerable efforts and resources have been devoted to adaptation methods for improving downstream performance of VLMs, particularly on parameter-efficient fine-tuning methods like prompt learning. However, a crucial aspect that has been largely overlooked is the confidence calibration problem in fine-tuned VLMs, which could greatly reduce reliability when deploying such models in the real world. This paper bridges the gap by systematically investigating the confidence calibration problem in the context of prompt learning and reveals that existing calibration methods are insufficient to address the problem, especially in the open-vocabulary setting. To solve the problem, we present a simple and effective approach called Distance-Aware Calibration (DAC), which is based on scaling the temperature using as guidance the distance between predicted text labels and base classes. The experiments with 7 distinct prompt learning methods applied across 11 diverse downstream datasets demonstrate the effectiveness of DAC, which achieves high efficacy without sacrificing the inference speed.

In tasks aiming for long-term returns, planning becomes necessary. We study generative modeling for planning with datasets repurposed from offline reinforcement learning. Specifically, we identify temporal consistency in the absence of step-wise rewards as one key technical challenge. We introduce the Latent Plan Transformer (LPT), a novel model that leverages a latent space to connect a Transformer-based trajectory generator and the final return. LPT can be learned with maximum likelihood estimation on trajectory-return pairs. In learning, posterior sampling of the latent variable naturally gathers sub-trajectories to form a consistent abstraction despite the finite context. During test time, the latent variable is inferred from an expected return before policy execution, realizing the idea of planning as inference. It then guides the autoregressive policy throughout the episode, functioning as a plan. Our experiments demonstrate that LPT can discover improved decisions from suboptimal trajectories. It achieves competitive performance across several benchmarks, including Gym-Mujoco, Maze2D, and Connect Four, exhibiting capabilities of nuanced credit assignments, trajectory stitching, and adaptation to environmental contingencies. These results validate that latent variable inference can be a strong alternative to step-wise reward prompting.

How would randomly shuffling feature vectors among nodes from the same class affect graph neural networks (GNNs)? The feature shuffle, intuitively, perturbs the dependence between graph topology and features (A-X dependence) for GNNs to learn from. Surprisingly, we observe a consistent and significant improvement in GNN performance following the feature shuffle. Having overlooked the impact of A-X dependence on GNNs, the prior literature does not provide a satisfactory understanding of the phenomenon. Thus, we raise two research questions. First, how should A-X dependence be measured, while controlling for potential confounds? Second, how does A-X dependence affect GNNs? In response, we (i) propose a principled measure for A-X dependence, (ii) design a random graph model that controls A-X dependence, (iii) establish a theory on how A-X dependence relates to graph convolution, and (iv) present empirical analysis on real-world graphs that aligns with the theory. We conclude that A-X dependence mediates the effect of graph convolution, such that smaller dependence improves GNN-based node classification.

Video sequences exhibit significant nuisance the-art results on NTU-60, NTU-120, Kinetics-skeleton and variations (undesired effects) of speed of actions, temporal UWA3D Multiview Activity II on supervised and unsupervised locations, and subjects' poses, leading to temporalviewpoint FSAR, and their meta-learning inspired fusion. Thus, we propose Joint tEmporal and cAmera viewpoiNt alIgnmEnt 1 Introduction (JEANIE) for sequence pairs. In particular, we focus on 3D skeleton sequences whose camera and subjects' poses can be Action recognition is a key topic in computer vision, easily manipulated in 3D. We evaluate JEANIE on skeletal with applications in video surveillance [105, 109, 120], Few-shot Action Recognition (FSAR), where matching well human-computer interaction, sport analysis and robotics. Given a query sequence, we create its several views labeling videos for 3D skeleton sequences is laborious, and by simulating several camera locations. For a support sequence, such pipelines need to be retrained or finetuned for new class we match it with view-simulated query sequences, concepts. Specifically, two-stream neural network [24, 23, 124] and 3D Convolutional each support temporal block can be matched to the Neural Network (3D CNN) [99, 9] aggregate framewise query temporal block with the same or adjacent (next) temporal and temporal block representations, respectively. However, index, and adjacent camera views to achieve joint local such networks are trained on large-scale datasets such temporal-viewpoint warping. JEANIE selects the smallest as Kinetics [9, 116, 110, 118] under a fixed set of training distance among matching paths with different temporalviewpoint classes. We also propose an Few-shot Learning (FSL) models for action recognition, unsupervised FSAR akin to clustering of sequences with termed Few-shot Action Recognition (FSAR), that rapidly JEANIE as a distance measure. JEANIE achieves state-of-adapt to novel classes given few training samples [77, 129, 31, 19, 138, 7, 112]. L. Wang is a Research Fellow at the School of Computing, the Australian J. Liu is an Assistant Professor at the Singapore University of Technology L. Zheng is an Associate Professor in the School of Computing, ANU.

In this paper, we consider a decentralized learning problem in the presence of stragglers. Although gradient coding techniques have been developed for distributed learning to evade stragglers, where the devices send encoded gradients with redundant training data, it is difficult to apply those techniques directly to decentralized learning scenarios. To deal with this problem, we propose a new gossip-based decentralized learning method with gradient coding (GOCO). In the proposed method, to avoid the negative impact of stragglers, the parameter vectors are updated locally using encoded gradients based on the framework of stochastic gradient coding and then averaged in a gossip-based manner. We analyze the convergence performance of GOCO for strongly convex loss functions. And we also provide simulation results to demonstrate the superiority of the proposed method in terms of learning performance compared with the baseline methods.

Modeling the correlations among errors is closely associated with how accurately the model can quantify predictive uncertainty in probabilistic time series forecasting. Recent multivariate models have made significant progress in accounting for contemporaneous correlations among errors, while a common assumption on these errors is that they are temporally independent for the sake of statistical simplicity. However, real-world observations often deviate from this assumption, since errors usually exhibit substantial autocorrelation due to various factors such as the exclusion of temporally correlated covariates. In this work, we propose an efficient method, based on a low-rank-plus-diagonal parameterization of the covariance matrix, which can effectively characterize the autocorrelation of errors. The proposed method possesses several desirable properties: the complexity does not scale with the number of time series, the resulting covariance can be used for calibrating predictions, and it can seamlessly integrate with any model with Gaussian-distributed errors. We empirically demonstrate these properties using two distinct neural forecasting models-GPVar and Transformer. Our experimental results confirm the effectiveness of our method in enhancing predictive accuracy and the quality of uncertainty quantification on multiple real-world datasets.

Linear discriminant analysis (LDA) is a widely used technique for data classification. The method offers adequate performance in many classification problems, but it becomes inefficient when the data covariance matrix is ill-conditioned. This often occurs when the feature space's dimensionality is higher than or comparable to the training data size. Regularized LDA (RLDA) methods based on regularized linear estimators of the data covariance matrix have been proposed to cope with such a situation. The performance of RLDA methods is well studied, with optimal regularization schemes already proposed. In this paper, we investigate the capability of a positive semidefinite ridge-type estimator of the inverse covariance matrix that coincides with a nonlinear (NL) covariance matrix estimator. The estimator is derived by reformulating the score function of the optimal classifier utilizing linear estimation methods, which eventually results in the proposed NL-RLDA classifier. We derive asymptotic and consistent estimators of the proposed technique's misclassification rate under the assumptions of a double-asymptotic regime and multivariate Gaussian model for the classes. The consistent estimator, coupled with a one-dimensional grid search, is used to set the value of the regularization parameter required for the proposed NL-RLDA classifier. Performance evaluations based on both synthetic and real data demonstrate the effectiveness of the proposed classifier. The proposed technique outperforms state-of-art methods over multiple datasets. When compared to state-of-the-art methods across various datasets, the proposed technique exhibits superior performance.

Machine unlearning has raised significant interest with the adoption of laws ensuring the ``right to be forgotten''. Researchers have provided a probabilistic notion of approximate unlearning under a similar definition of Differential Privacy (DP), where privacy is defined as statistical indistinguishability to retraining from scratch. We propose Langevin unlearning, an unlearning framework based on noisy gradient descent with privacy guarantees for approximate unlearning problems. Langevin unlearning unifies the DP learning process and the privacy-certified unlearning process with many algorithmic benefits. These include approximate certified unlearning for non-convex problems, complexity saving compared to retraining, sequential and batch unlearning for multiple unlearning requests. We verify the practicality of Langevin unlearning by studying its privacy-utility-complexity trade-off via experiments on benchmark datasets, and also demonstrate its superiority against gradient-decent-plus-output-perturbation based approximate unlearning.

Spatio-temporal forecasting, pivotal in numerous fields, hinges on the delicate equilibrium between isolating nuanced patterns and sifting out noise. To tackle this, we introduce Sparse Regression-based Vector Quantization (SVQ), a novel technique that leverages sparse regression for succinct representation, an approach theoretically and practically favored over classical clustering-based vector quantization methods. This approach preserves critical details from the original vectors using a regression model while filtering out noise via sparse design. Moreover, we approximate the sparse regression process using a blend of a two-layer MLP and an extensive codebook. This approach not only substantially cuts down on computational costs but also grants SVQ differentiability and training simplicity, resulting in a notable enhancement of performance. Our empirical studies on five spatial-temporal benchmark datasets demonstrate that SVQ achieves state-of-the-art results. Specifically, on the WeatherBench-S temperature dataset, SVQ improves the top baseline by 7.9%. In video prediction benchmarks-Human, KTH, and KittiCaltech-it reduces MAE by an average of 9.4% and improves image quality by 17.3% (LPIPS).

Human Activity Recognition (HAR) models often suffer from performance degradation in real-world applications due to distribution shifts in activity patterns across individuals. Test-Time Adaptation (TTA) is an emerging learning paradigm that aims to utilize the test stream to adjust predictions in real-time inference, which has not been explored in HAR before. However, the high computational cost of optimization-based TTA algorithms makes it intractable to run on resource-constrained edge devices. In this paper, we propose an Optimization-Free Test-Time Adaptation (OFTTA) framework for sensor-based HAR. OFTTA adjusts the feature extractor and linear classifier simultaneously in an optimization-free manner. For the feature extractor, we propose Exponential DecayTest-time Normalization (EDTN) to replace the conventional batch normalization (CBN) layers. EDTN combines CBN and Test-time batch Normalization (TBN) to extract reliable features against domain shifts with TBN's influence decreasing exponentially in deeper layers. For the classifier, we adjust the prediction by computing the distance between the feature and the prototype, which is calculated by a maintained support set. In addition, the update of the support set is based on the pseudo label, which can benefit from reliable features extracted by EDTN. Extensive experiments on three public cross-person HAR datasets and two different TTA settings demonstrate that OFTTA outperforms the state-of-the-art TTA approaches in both classification performance and computational efficiency. Finally, we verify the superiority of our proposed OFTTA on edge devices, indicating possible deployment in real applications. Our code is available at https://github.com/Claydon-Wang/OFTTA.