Classical convergence analyses for optimization algorithms rely on the widely-adopted uniform smoothness assumption. However, recent experimental studies have demonstrated that many machine learning problems exhibit non-uniform smoothness, meaning the smoothness factor is a function of the model parameter instead of a universal constant. In particular, it has been observed that the smoothness grows with respect to the gradient norm along the training trajectory. Motivated by this phenomenon, the recently introduced $(L_0, L_1)$-smoothness is a more general notion, compared to traditional $L$-smoothness, that captures such positive relationship between smoothness and gradient norm. Under this type of non-uniform smoothness, existing literature has designed stochastic first-order algorithms by utilizing gradient clipping techniques to obtain the optimal $\mathcal{O}(\epsilon^{-3})$ sample complexity for finding an $\epsilon$-approximate first-order stationary solution. Nevertheless, the studies of quasi-Newton methods are still lacking. Considering higher accuracy and more robustness for quasi-Newton methods, in this paper we propose a fast stochastic quasi-Newton method when there exists non-uniformity in smoothness. Leveraging gradient clipping and variance reduction, our algorithm can achieve the best-known $\mathcal{O}(\epsilon^{-3})$ sample complexity and enjoys convergence speedup with simple hyperparameter tuning. Our numerical experiments show that our proposed algorithm outperforms the state-of-the-art approaches.

Keywords Abstract Developing algorithms for accurate and comprehensive neural decoding of mental contents is one of the longcherished goals in the field of neuroscience and brain-machine interfaces. Previous studies have demonstrated the feasibility of neural decoding by training machine learning models to map brain activity patterns into a semantic vector representation of stimuli. These vectors, hereafter referred as pretrained feature vectors, are usually derived from semantic spaces based solely on image and/or text features and therefore they might have a totally different characteristics than how visual stimuli is represented in the human brain, resulting in limiting the capability of brain decoders to learn this mapping. To address this issue, we propose a representation learning framework, termed brain-grounding of semantic vectors, which fine-tunes pretrained feature vectors to better align with the neural representation of visual stimuli in the human brain. We trained this model this model with functional magnetic resonance imaging (fMRI) of 150 different visual stimuli categories, and then performed zero-shot brain decoding and identification analyses on 1) fMRI and 2) magnetoencephalography (MEG). Interestingly, we observed that by using the brain-grounded vectors, the brain decoding and identification accuracy on brain data from different neuroimaging modalities increases. These findings underscore the potential of incorporating a richer array of brain-derived features to enhance performance of brain decoding algorithms.

Existing example-based prediction explanation methods often bridge test and training data points through the model's parameters or latent representations. While these methods offer clues to the causes of model predictions, they often exhibit innate shortcomings, such as incurring significant computational overhead or producing coarse-grained explanations. This paper presents a Highly-precise and Data-centric Explanation (HD-Explain), a straightforward prediction explanation method exploiting properties of Kernelized Stein Discrepancy (KSD). Specifically, the KSD uniquely defines a parameterized kernel function for a trained model that encodes model-dependent data correlation. By leveraging the kernel function, one can identify training samples that provide the best predictive support to a test point efficiently. We conducted thorough analyses and experiments across multiple classification domains, where we show that HD-Explain outperforms existing methods from various aspects, including 1) preciseness (fine-grained explanation), 2) consistency, and 3) computation efficiency, leading to a surprisingly simple, effective, and robust prediction explanation solution.

Efforts to leverage deep learning models in low-resource regimes have led to numerous augmentation studies. However, the direct application of methods such as mixup and cutout to text data, is limited due to their discrete characteristics. While methods using pretrained language models have exhibited efficiency, they require additional considerations for robustness. Inspired by recent studies on decision boundaries, this paper proposes a decision-boundary-aware data augmentation strategy to enhance robustness using pretrained language models. The proposed technique first focuses on shifting the latent features closer to the decision boundary, followed by reconstruction to generate an ambiguous version with a soft label. Additionally, mid-K sampling is suggested to enhance the diversity of the generated sentences. This paper demonstrates the performance of the proposed augmentation strategy compared to other methods through extensive experiments. Furthermore, the ablation study reveals the effect of soft labels and mid-K sampling and the extensibility of the method with curriculum data augmentation.

This paper aims to clearly distinguish between Stochastic Gradient Descent with Momentum (SGDM) and Adam in terms of their convergence rates. We demonstrate that Adam achieves a faster convergence compared to SGDM under the condition of non-uniformly bounded smoothness. Our findings reveal that: (1) in deterministic environments, Adam can attain the known lower bound for the convergence rate of deterministic first-order optimizers, whereas the convergence rate of Gradient Descent with Momentum (GDM) has higher order dependence on the initial function value; (2) in stochastic setting, Adam's convergence rate upper bound matches the lower bounds of stochastic first-order optimizers, considering both the initial function value and the final error, whereas there are instances where SGDM fails to converge with any learning rate. These insights distinctly differentiate Adam and SGDM regarding their convergence rates. Additionally, by introducing a novel stopping-time based technique, we further prove that if we consider the minimum gradient norm during iterations, the corresponding convergence rate can match the lower bounds across all problem hyperparameters. The technique can also help proving that Adam with a specific hyperparameter scheduler is parameter-agnostic, which hence can be of independent interest.

In recent several years, the information bottleneck (IB) principle provides an information-theoretic framework for deep multi-view clustering (MVC) by compressing multi-view observations while preserving the relevant information of multiple views. Although existing IB-based deep MVC methods have achieved huge success, they rely on variational approximation and distribution assumption to estimate the lower bound of mutual information, which is a notoriously hard and impractical problem in high-dimensional multi-view spaces. In this work, we propose a new differentiable information bottleneck (DIB) method, which provides a deterministic and analytical MVC solution by fitting the mutual information without the necessity of variational approximation. Specifically, we first propose to directly fit the mutual information of high-dimensional spaces by leveraging normalized kernel Gram matrix, which does not require any auxiliary neural estimator to estimate the lower bound of mutual information. Then, based on the new mutual information measurement, a deterministic multi-view neural network with analytical gradients is explicitly trained to parameterize IB principle, which derives a deterministic compression of input variables from different views. Finally, a triplet consistency discovery mechanism is devised, which is capable of mining the feature consistency, cluster consistency and joint consistency based on the deterministic and compact representations. Extensive experimental results show the superiority of our DIB method on 6 benchmarks compared with 13 state-of-the-art baselines.

Graph Neural Networks (GNN) have emerged as a popular and standard approach for learning from graph-structured data. The literature on GNN highlights the potential of this evolving research area and its widespread adoption in real-life applications. However, most of the approaches are either new in concept or derived from specific techniques. Therefore, the potential of more than one approach in hybrid form has not been studied extensively, which can be well utilized for sequenced data or static data together. We derive a hybrid approach based on two established techniques as generalized aggregation networks and topology adaptive graph convolution networks that solve our purpose to apply on both types of sequenced and static nature of data, effectively. The proposed method applies to both node and graph classification. Our empirical analysis reveals that the results are at par with literature results and better for handwritten strokes as sequenced data, where graph structures have not been explored.

Manually annotating datasets for training deep models is very labor-intensive and time-consuming. To overcome such inferiority, directly leveraging web images to conduct training data becomes a natural choice. Nevertheless, the presence of label noise in web data usually degrades the model performance. Existing methods for combating label noise are typically designed and tested on synthetic noisy datasets. However, they tend to fail to achieve satisfying results on real-world noisy datasets. To this end, we propose a method named GRIP to alleviate the noisy label problem for both synthetic and real-world datasets. Specifically, GRIP utilizes a group regularization strategy that estimates class soft labels to improve noise robustness. Soft label supervision reduces overfitting on noisy labels and learns inter-class similarities to benefit classification. Furthermore, an instance purification operation globally identifies noisy labels by measuring the difference between each training sample and its class soft label. Through operations at both group and instance levels, our approach integrates the advantages of noise-robust and noise-cleaning methods and remarkably alleviates the performance degradation caused by noisy labels. Comprehensive experimental results on synthetic and real-world datasets demonstrate the superiority of GRIP over the existing state-of-the-art methods.

Decision making and learning in the presence of uncertainty has attracted significant attention in view of the increasing need to achieve robust and reliable operations. In the case where uncertainty stems from the presence of adversarial attacks this need is becoming more prominent. In this paper we focus on linear and nonlinear classification problems and propose a novel adversarial training method for robust classifiers, inspired by Support Vector Machine (SVM) margins. We view robustness under a data driven lens, and derive finite sample complexity bounds for both linear and non-linear classifiers in binary and multi-class scenarios. Notably, our bounds match natural classifiers' complexity. Our algorithm minimizes a worst-case surrogate loss using Linear Programming (LP) and Second Order Cone Programming (SOCP) for linear and non-linear models. Numerical experiments on the benchmark MNIST and CIFAR10 datasets show our approach's comparable performance to state-of-the-art methods, without needing adversarial examples during training. Our work offers a comprehensive framework for enhancing binary linear and non-linear classifier robustness, embedding robustness in learning under the presence of adversaries.

Self-supervised learning (SSL) has proven effective in learning high-quality representations for various downstream tasks, with a primary focus on semantic tasks. However, its application in geometric tasks remains underexplored, partially due to the absence of a standardized evaluation method for geometric representations. To address this gap, we introduce a new pose-estimation benchmark for assessing SSL geometric representations, which demands training without semantic or pose labels and achieving proficiency in both semantic and geometric downstream tasks. On this benchmark, we study enhancing SSL geometric representations without sacrificing semantic classification accuracy. We find that leveraging mid-layer representations improves pose-estimation performance by 10-20%. Further, we introduce an unsupervised trajectory-regularization loss, which improves performance by an additional 4% and improves generalization ability on out-of-distribution data. We hope the proposed benchmark and methods offer new insights and improvements in self-supervised geometric representation learning.