Bilevel reinforcement learning via the development of hyper-gradient without lower-level convexity

Bilevel reinforcement learning (RL), which features intertwined two-level problems, has attracted growing interest recently. The inherent non-convexity of the lower-level RL problem is, however, to be an impediment to developing bilevel optimization methods. By employing the fixed point equation associated with the regularized RL, we characterize the hyper-gradient via fully first-order information, thus circumventing the assumption of lower-level convexity. This, remarkably, distinguishes our development of hyper-gradient from the general AID-based bilevel frameworks since we take advantage of the specific structure of RL problems. Moreover, we propose both model-based and model-free bilevel reinforcement learning algorithms, facilitated by access to the fully first-order hyper-gradient. Both algorithms are provable to enjoy the convergence rate $\mathcal{O}(\epsilon^{-1})$. To the best of our knowledge, this is the first time that AID-based bilevel RL gets rid of additional assumptions on the lower-level problem. In addition, numerical experiments demonstrate that the hyper-gradient indeed serves as an integration of exploitation and exploration.

Optimization without retraction on the random generalized Stiefel manifold

Optimization over the set of matrices that satisfy $X^\top B X = I_p$, referred to as the generalized Stiefel manifold, appears in many applications involving sampled covariance matrices such as canonical correlation analysis (CCA), independent component analysis (ICA), and the generalized eigenvalue problem (GEVP). Solving these problems is typically done by iterative methods, such as Riemannian approaches, which require a computationally expensive eigenvalue decomposition involving fully formed $B$. We propose a cheap stochastic iterative method that solves the optimization problem while having access only to a random estimate of the feasible set. Our method does not enforce the constraint in every iteration exactly, but instead it produces iterations that converge to a critical point on the generalized Stiefel manifold defined in expectation. The method has lower per-iteration cost, requires only matrix multiplications, and has the same convergence rates as its Riemannian counterparts involving the full matrix $B$. Experiments demonstrate its effectiveness in various machine learning applications involving generalized orthogonality constraints, including CCA, ICA, and GEVP.

LancBiO: dynamic Lanczos-aided bilevel optimization via Krylov subspace

Bilevel optimization, with broad applications in machine learning, has an intricate hierarchical structure. Gradient-based methods have emerged as a common approach to large-scale bilevel problems. However, the computation of the hyper-gradient, which involves a Hessian inverse vector product, confines the efficiency and is regarded as a bottleneck. To circumvent the inverse, we construct a sequence of low-dimensional approximate Krylov subspaces with the aid of the Lanczos process. As a result, the constructed subspace is able to dynamically and incrementally approximate the Hessian inverse vector product with less effort and thus leads to a favorable estimate of the hyper-gradient. Moreover, we propose a~provable subspace-based framework for bilevel problems where one central step is to solve a small-size tridiagonal linear system. To the best of our knowledge, this is the first time that subspace techniques are incorporated into bilevel optimization. This successful trial not only enjoys $\mathcal{O}(\epsilon^{-1})$ convergence rate but also demonstrates efficiency in a synthetic problem and two deep learning tasks.

AttentionStore: Cost-effective Attention Reuse across Multi-turn Conversations in Large Language Model Serving

Interacting with humans through multi-turn conversations is a fundamental feature of large language models (LLMs). However, existing LLM serving engines for executing multi-turn conversations are inefficient due to the need to repeatedly compute the key-value (KV) caches of historical tokens, incurring high serving costs. To address the problem, this paper proposes AttentionStore, a new attention mechanism that enables the reuse of KV caches (i.e., attention reuse) across multi-turn conversations, significantly reducing the repetitive computation overheads. AttentionStore maintains a hierarchical KV caching system that leverages cost-effective memory/storage mediums to save KV caches for all requests. To reduce KV cache access overheads from slow mediums, AttentionStore employs layer-wise pre-loading and asynchronous saving schemes to overlap the KV cache access with the GPU computation. To ensure that the KV caches to be accessed are placed in the fastest hierarchy, AttentionStore employs scheduler-aware fetching and eviction schemes to consciously place the KV caches in different layers based on the hints from the inference job scheduler. To avoid the invalidation of the saved KV caches incurred by context window overflow, AttentionStore enables the saved KV caches to remain valid via decoupling the positional encoding and effectively truncating the KV caches. Extensive experimental results demonstrate that AttentionStore significantly decreases the time to the first token (TTFT) by up to 88%, improves the prompt prefilling throughput by 8.2$\times$ for multi-turn conversations, and reduces the end-to-end inference cost by up to 56%. For long sequence inference, AttentionStore reduces the TTFT by up to 95% and improves the prompt prefilling throughput by 22$\times$.

Forging Vision Foundation Models for Autonomous Driving: Challenges, Methodologies, and Opportunities

The rise of large foundation models, trained on extensive datasets, is revolutionizing the field of AI. Models such as SAM, DALL-E2, and GPT-4 showcase their adaptability by extracting intricate patterns and performing effectively across diverse tasks, thereby serving as potent building blocks for a wide range of AI applications. Autonomous driving, a vibrant front in AI applications, remains challenged by the lack of dedicated vision foundation models (VFMs). The scarcity of comprehensive training data, the need for multi-sensor integration, and the diverse task-specific architectures pose significant obstacles to the development of VFMs in this field. This paper delves into the critical challenge of forging VFMs tailored specifically for autonomous driving, while also outlining future directions. Through a systematic analysis of over 250 papers, we dissect essential techniques for VFM development, including data preparation, pre-training strategies, and downstream task adaptation. Moreover, we explore key advancements such as NeRF, diffusion models, 3D Gaussian Splatting, and world models, presenting a comprehensive roadmap for future research. To empower researchers, we have built and maintained https://github.com/zhanghm1995/Forge_VFM4AD, an open-access repository constantly updated with the latest advancements in forging VFMs for autonomous driving.

Segment Anything in Defect Detection

Defect detection plays a crucial role in infrared non-destructive testing systems, offering non-contact, safe, and efficient inspection capabilities. However, challenges such as low resolution, high noise, and uneven heating in infrared thermal images hinder comprehensive and accurate defect detection. In this study, we propose DefectSAM, a novel approach for segmenting defects on highly noisy thermal images based on the widely adopted model, Segment Anything (SAM)\cite{kirillov2023segany}. Harnessing the power of a meticulously curated dataset generated through labor-intensive lab experiments and valuable prompts from experienced experts, DefectSAM surpasses existing state-of-the-art segmentation algorithms and achieves significant improvements in defect detection rates. Notably, DefectSAM excels in detecting weaker and smaller defects on complex and irregular surfaces, reducing the occurrence of missed detections and providing more accurate defect size estimations. Experimental studies conducted on various materials have validated the effectiveness of our solutions in defect detection, which hold significant potential to expedite the evolution of defect detection tools, enabling enhanced inspection capabilities and accuracy in defect identification.

Infeasible Deterministic, Stochastic, and Variance-Reduction Algorithms for Optimization under Orthogonality Constraints

Orthogonality constraints naturally appear in many machine learning problems, from Principal Components Analysis to robust neural network training. They are usually solved using Riemannian optimization algorithms, which minimize the objective function while enforcing the constraint. However, enforcing the orthogonality constraint can be the most time-consuming operation in such algorithms. Recently, Ablin & Peyr\'e (2022) proposed the Landing algorithm, a method with cheap iterations that does not enforce the orthogonality constraint but is attracted towards the manifold in a smooth manner. In this article, we provide new practical and theoretical developments for the landing algorithm. First, the method is extended to the Stiefel manifold, the set of rectangular orthogonal matrices. We also consider stochastic and variance reduction algorithms when the cost function is an average of many functions. We demonstrate that all these methods have the same rate of convergence as their Riemannian counterparts that exactly enforce the constraint. Finally, our experiments demonstrate the promise of our approach to an array of machine-learning problems that involve orthogonality constraints.

UCC: Uncertainty guided Cross-head Co-training for Semi-Supervised Semantic Segmentation

Deep neural networks (DNNs) have witnessed great successes in semantic segmentation, which requires a large number of labeled data for training. We present a novel learning framework called Uncertainty guided Cross-head Co-training (UCC) for semi-supervised semantic segmentation. Our framework introduces weak and strong augmentations within a shared encoder to achieve co-training, which naturally combines the benefits of consistency and self-training. Every segmentation head interacts with its peers and, the weak augmentation result is used for supervising the strong. The consistency training samples' diversity can be boosted by Dynamic Cross-Set Copy-Paste (DCSCP), which also alleviates the distribution mismatch and class imbalance problems. Moreover, our proposed Uncertainty Guided Re-weight Module (UGRM) enhances the self-training pseudo labels by suppressing the effect of the low-quality pseudo labels from its peer via modeling uncertainty. Extensive experiments on Cityscapes and PASCAL VOC 2012 demonstrate the effectiveness of our UCC. Our approach significantly outperforms other state-of-the-art semi-supervised semantic segmentation methods. It achieves 77.17$\%$, 76.49$\%$ mIoU on Cityscapes and PASCAL VOC 2012 datasets respectively under 1/16 protocols, which are +10.1$\%$, +7.91$\%$ better than the supervised baseline.

Brachial Plexus Nerve Trunk Segmentation Using Deep Learning: A Comparative Study with Doctors' Manual Segmentation

Ultrasound-guided nerve block anesthesia (UGNB) is a high-tech visual nerve block anesthesia method that can observe the target nerve and its surrounding structures, the puncture needle's advancement, and local anesthetics spread in real-time. The key in UGNB is nerve identification. With the help of deep learning methods, the automatic identification or segmentation of nerves can be realized, assisting doctors in completing nerve block anesthesia accurately and efficiently. Here, we establish a public dataset containing 320 ultrasound images of brachial plexus (BP). Three experienced doctors jointly produce the BP segmentation ground truth and label brachial plexus trunks. We design a brachial plexus segmentation system (BPSegSys) based on deep learning. BPSegSys achieves experienced-doctor-level nerve identification performance in various experiments. We evaluate BPSegSys' performance in terms of intersection-over-union (IoU), a commonly used performance measure for segmentation experiments. Considering three dataset groups in our established public dataset, the IoU of BPSegSys are 0.5238, 0.4715, and 0.5029, respectively, which exceed the IoU 0.5205, 0.4704, and 0.4979 of experienced doctors. In addition, we show that BPSegSys can help doctors identify brachial plexus trunks more accurately, with IoU improvement up to 27%, which has significant clinical application value.

EAGAN: Efficient Two-stage Evolutionary Architecture Search for GANs

Generative Adversarial Networks (GANs) have been proven hugely successful in image generation tasks, but GAN training has the problem of instability. Many works have improved the stability of GAN training by manually modifying the GAN architecture, which requires human expertise and extensive trial-and-error. Thus, neural architecture search (NAS), which aims to automate the model design, has been applied to search GANs on the task of unconditional image generation. The early NAS-GAN works only search generators for reducing the difficulty. Some recent works have attempted to search both generator (G) and discriminator (D) to improve GAN performance, but they still suffer from the instability of GAN training during the search. To alleviate the instability issue, we propose an efficient two-stage evolutionary algorithm (EA) based NAS framework to discover GANs, dubbed \textbf{EAGAN}. Specifically, we decouple the search of G and D into two stages and propose the weight-resetting strategy to improve the stability of GAN training. Besides, we perform evolution operations to produce the Pareto-front architectures based on multiple objectives, resulting in a superior combination of G and D. By leveraging the weight-sharing strategy and low-fidelity evaluation, EAGAN can significantly shorten the search time. EAGAN achieves highly competitive results on the CIFAR-10 (IS=8.81$\pm$0.10, FID=9.91) and surpasses previous NAS-searched GANs on the STL-10 dataset (IS=10.44$\pm$0.087, FID=22.18).