As deep neural networks (DNNs) are being applied to a wide range of edge intelligent applications, it is critical for edge inference platforms to have both high-throughput and low-latency at the same time. Such edge platforms with multiple DNN models pose new challenges for scheduler designs. First, each request may have different service level objectives (SLOs) to improve quality of service (QoS). Second, the edge platforms should be able to efficiently schedule multiple heterogeneous DNN models so that system utilization can be improved. To meet these two goals, this paper proposes BCEdge, a novel learning-based scheduling framework that takes adaptive batching and concurrent execution of DNN inference services on edge platforms. We define a utility function to evaluate the trade-off between throughput and latency. The scheduler in BCEdge leverages maximum entropy-based deep reinforcement learning (DRL) to maximize utility by 1) co-optimizing batch size and 2) the number of concurrent models automatically. Our prototype implemented on different edge platforms shows that the proposed BCEdge enhances utility by up to 37.6% on average, compared to state-of-the-art solutions, while satisfying SLOs.
Correlated time series (CTS) forecasting plays an essential role in many practical applications, such as traffic management and server load control. Many deep learning models have been proposed to improve the accuracy of CTS forecasting. However, while models have become increasingly complex and computationally intensive, they struggle to improve accuracy. Pursuing a different direction, this study aims instead to enable much more efficient, lightweight models that preserve accuracy while being able to be deployed on resource-constrained devices. To achieve this goal, we characterize popular CTS forecasting models and yield two observations that indicate directions for lightweight CTS forecasting. On this basis, we propose the LightCTS framework that adopts plain stacking of temporal and spatial operators instead of alternate stacking that is much more computationally expensive. Moreover, LightCTS features light temporal and spatial operator modules, called L-TCN and GL-Former, that offer improved computational efficiency without compromising their feature extraction capabilities. LightCTS also encompasses a last-shot compression scheme to reduce redundant temporal features and speed up subsequent computations. Experiments with single-step and multi-step forecasting benchmark datasets show that LightCTS is capable of nearly state-of-the-art accuracy at much reduced computational and storage overheads.
Adaptive gradient algorithms borrow the moving average idea of heavy ball acceleration to estimate accurate first- and second-order moments of gradient for accelerating convergence. However, Nesterov acceleration which converges faster than heavy ball acceleration in theory and also in many empirical cases is much less investigated under the adaptive gradient setting. In this work, we propose the ADAptive Nesterov momentum algorithm, Adan for short, to speed up the training of deep neural networks effectively. Adan first reformulates the vanilla Nesterov acceleration to develop a new Nesterov momentum estimation (NME) method, which avoids the extra computation and memory overhead of computing gradient at the extrapolation point. Then Adan adopts NME to estimate the first- and second-order moments of the gradient in adaptive gradient algorithms for convergence acceleration. Besides, we prove that Adan finds an $\epsilon$-approximate first-order stationary point within $O(\epsilon^{-3.5})$ stochastic gradient complexity on the nonconvex stochastic problems (e.g., deep learning problems), matching the best-known lower bound. Extensive experimental results show that Adan surpasses the corresponding SoTA optimizers on both vision transformers (ViTs) and CNNs, and sets new SoTAs for many popular networks, e.g., ResNet, ConvNext, ViT, Swin, MAE, LSTM, Transformer-XL, and BERT. More surprisingly, Adan can use half of the training cost (epochs) of SoTA optimizers to achieve higher or comparable performance on ViT and ResNet, e.t.c., and also shows great tolerance to a large range of minibatch size, e.g., from 1k to 32k. We hope Adan can contribute to the development of deep learning by reducing training cost and relieving engineering burden of trying different optimizers on various architectures. Code is released at https://github.com/sail-sg/Adan.
Hierarchical clustering over graphs is a fundamental task in data mining and machine learning with applications in domains such as phylogenetics, social network analysis, and information retrieval. Specifically, we consider the recently popularized objective function for hierarchical clustering due to Dasgupta. Previous algorithms for (approximately) minimizing this objective function require linear time/space complexity. In many applications the underlying graph can be massive in size making it computationally challenging to process the graph even using a linear time/space algorithm. As a result, there is a strong interest in designing algorithms that can perform global computation using only sublinear resources. The focus of this work is to study hierarchical clustering for massive graphs under three well-studied models of sublinear computation which focus on space, time, and communication, respectively, as the primary resources to optimize: (1) (dynamic) streaming model where edges are presented as a stream, (2) query model where the graph is queried using neighbor and degree queries, (3) MPC model where the graph edges are partitioned over several machines connected via a communication channel. We design sublinear algorithms for hierarchical clustering in all three models above. At the heart of our algorithmic results is a view of the objective in terms of cuts in the graph, which allows us to use a relaxed notion of cut sparsifiers to do hierarchical clustering while introducing only a small distortion in the objective function. Our main algorithmic contributions are then to show how cut sparsifiers of the desired form can be efficiently constructed in the query model and the MPC model. We complement our algorithmic results by establishing nearly matching lower bounds that rule out the possibility of designing better algorithms in each of these models.
This paper studies the accelerated gradient descent for general nonconvex problems under the gradient Lipschitz and Hessian Lipschitz assumptions. We establish that a simple restarted accelerated gradient descent (AGD) finds an $\epsilon$-approximate first-order stationary point in $O(\epsilon^{-7/4})$ gradient computations with simple proofs. Our complexity does not hide any polylogarithmic factors, and thus it improves over the state-of-the-art one by the $O(\log\frac{1}{\epsilon})$ factor. Our simple algorithm only consists of Nesterov's classical AGD and a restart mechanism, and it does not need the negative curvature exploitation or the optimization of regularized surrogate functions. Technically, our simple proof does not invoke the analysis for the strongly convex AGD, which is crucial to remove the $O(\log\frac{1}{\epsilon})$ factor.
Decentralized optimization over time-varying graphs has been increasingly common in modern machine learning with massive data stored on millions of mobile devices, such as in federated learning. This paper revisits the widely used accelerated gradient tracking and extends it to time-varying graphs. We prove the $O((\frac{\gamma}{1-\sigma_{\gamma}})^2\sqrt{\frac{L}{\epsilon}})$ and $O((\frac{\gamma}{1-\sigma_{\gamma}})^{1.5}\sqrt{\frac{L}{\mu}}\log\frac{1}{\epsilon})$ complexities for the practical single loop accelerated gradient tracking over time-varying graphs when the problems are nonstrongly convex and strongly convex, respectively, where $\gamma$ and $\sigma_{\gamma}$ are two common constants charactering the network connectivity, $\epsilon$ is the desired precision, and $L$ and $\mu$ are the smoothness and strong convexity constants, respectively. Our complexities improve significantly over the ones of $O(\frac{1}{\epsilon^{5/7}})$ and $O((\frac{L}{\mu})^{5/7}\frac{1}{(1-\sigma)^{1.5}}\log\frac{1}{\epsilon})$, respectively, which were proved in the original literature only for static graphs, where $\frac{1}{1-\sigma}$ equals $\frac{\gamma}{1-\sigma_{\gamma}}$ when the network is time-invariant. When combining with a multiple consensus subroutine, the dependence on the network connectivity constants can be further improved to $O(1)$ and $O(\frac{\gamma}{1-\sigma_{\gamma}})$ for the computation and communication complexities, respectively. When the network is static, by employing the Chebyshev acceleration, our complexities exactly match the lower bounds without hiding any poly-logarithmic factor for both nonstrongly convex and strongly convex problems.