Computing Research Repository

Wang, Hui-Po; Stich, Sebastian U.; He, Yang; Fritz, Mario

doi: 10.48550/arxiv.2110.05323pmid: N/A

Abstract:Federated learning is a powerful distributed learning scheme that allows numerous edge devices to collaboratively train a model without sharing their data. However, training is resource-intensive for edge devices, and limited network bandwidth is often the main bottleneck. Prior work often overcomes the constraints by condensing the models or messages into compact formats, e.g., by gradient compression or distillation. In contrast, we propose ProgFed, the first progressive training framework for efficient and effective federated learning. It inherently reduces computation and two-way communication costs while maintaining the strong performance of the final models. We theoretically prove that ProgFed converges at the same asymptotic rate as standard training on full models. Extensive results on a broad range of architectures, including CNNs (VGG, ResNet, ConvNets) and U-nets, and diverse tasks from simple classification to medical image segmentation show that our highly effective training approach saves up to $20\%$ computation and up to $63\%$ communication costs for converged models. As our approach is also complimentary to prior work on compression, we can achieve a wide range of trade-offs by combining these techniques, showing reduced communication of up to $50\times$ at only $0.1\%$ loss in utility. Code is available at this https URL.

Guo, Yunxiao; Long, Han; Duan, Xiaojun; Feng, Kaiyuan; Li, Maochu; Ma, Xiaying

doi: 10.48550/arxiv.2110.10522pmid: N/A

Abstract:As a popular Deep Reinforcement Learning (DRL) algorithm, Proximal Policy Optimization (PPO) has demonstrated remarkable efficacy in numerous complex tasks. According to the penalty mechanism in a surrogate, PPO can be classified into PPO with KL divergence (PPO-KL) and PPO with Clip (PPO-Clip). In this paper, we analyze the impact of asymmetry in KL divergence on PPO-KL and highlight that when this asymmetry is pronounced, it will misguide the improvement of the surrogate. To address this issue, we represent the PPO-KL in inner product form and demonstrate that the KL divergence is a Correntropy Induced Metric (CIM) in Euclidean space. Subsequently, we extend the PPO-KL to the Reproducing Kernel Hilbert Space (RKHS), redefine the inner products with RKHS, and propose the PPO-CIM algorithm. Moreover, this paper states that the PPO-CIM algorithm has a lower computation cost in policy gradient and proves that PPO-CIM can guarantee the new policy is within the trust region while the kernel satisfies some conditions. Finally, we design experiments based on six Mujoco continuous-action tasks to validate the proposed algorithm. The experimental results validate that the asymmetry of KL divergence can affect the policy improvement of PPO-KL and show that the PPO-CIM can perform better than both PPO-KL and PPO-Clip in most tasks.

Mimmo, Nicola; Carnevale, Guido; Testa, Andrea; Notarstefano, Giuseppe

doi: 10.48550/arxiv.2110.04234pmid: N/A

Abstract:In this paper, we deal with a network of agents that want to cooperatively minimize the sum of local cost functions depending on a common decision variable. We consider the challenging scenario in which objective functions are unknown and agents have only access to local measurements of their local functions. We propose a novel distributed algorithm that combines a recent gradient tracking policy with an extremum-seeking technique to estimate the global descent direction. The joint use of these two techniques results in a distributed optimization scheme that provides arbitrarily accurate solution estimates through the combination of Lyapunov and averaging analysis approaches with consensus theory. We perform numerical simulations in a personalized optimization framework to corroborate the theoretical results.

Cai, Zhongang; Zhang, Mingyuan; Ren, Jiawei; Wei, Chen; Ren, Daxuan; Lin, Zhengyu; Zhao, Haiyu; Yang, Lei; Loy, Chen Change; Liu, Ziwei

doi: 10.48550/arxiv.2110.07588pmid: N/A

Abstract:Image- and video-based 3D human recovery (i.e., pose and shape estimation) have achieved substantial progress. However, due to the prohibitive cost of motion capture, existing datasets are often limited in scale and diversity. In this work, we obtain massive human sequences by playing the video game with automatically annotated 3D ground truths. Specifically, we contribute GTA-Human, a large-scale 3D human dataset generated with the GTA-V game engine, featuring a highly diverse set of subjects, actions, and scenarios. More importantly, we study the use of game-playing data and obtain five major insights. First, game-playing data is surprisingly effective. A simple frame-based baseline trained on GTA-Human outperforms more sophisticated methods by a large margin. For video-based methods, GTA-Human is even on par with the in-domain training set. Second, we discover that synthetic data provides critical complements to the real data that is typically collected indoor. Our investigation into domain gap provides explanations for our data mixture strategies that are simple yet useful. Third, the scale of the dataset matters. The performance boost is closely related to the additional data available. A systematic study reveals the model sensitivity to data density from multiple key aspects. Fourth, the effectiveness of GTA-Human is also attributed to the rich collection of strong supervision labels (SMPL parameters), which are otherwise expensive to acquire in real datasets. Fifth, the benefits of synthetic data extend to larger models such as deeper convolutional neural networks (CNNs) and Transformers, for which a significant impact is also observed. We hope our work could pave the way for scaling up 3D human recovery to the real world. Homepage: this https URL

Wei, Chen-Yu; Dann, Christoph; Zimmert, Julian

doi: 10.48550/arxiv.2110.03580pmid: N/A

Abstract:We develop a model selection approach to tackle reinforcement learning with adversarial corruption in both transition and reward. For finite-horizon tabular MDPs, without prior knowledge on the total amount of corruption, our algorithm achieves a regret bound of $\widetilde{\mathcal{O}}(\min\{\frac{1}{\Delta}, \sqrt{T}\}+C)$ where $T$ is the number of episodes, $C$ is the total amount of corruption, and $\Delta$ is the reward gap between the best and the second-best policy. This is the first worst-case optimal bound achieved without knowledge of $C$, improving previous results of Lykouris et al. (2021); Chen et al. (2021); Wu et al. (2021). For finite-horizon linear MDPs, we develop a computationally efficient algorithm with a regret bound of $\widetilde{\mathcal{O}}(\sqrt{(1+C)T})$, and another computationally inefficient one with $\widetilde{\mathcal{O}}(\sqrt{T}+C)$, improving the result of Lykouris et al. (2021) and answering an open question by Zhang et al. (2021b). Finally, our model selection framework can be easily applied to other settings including linear bandits, linear contextual bandits, and MDPs with general function approximation, leading to several improved or new results.

Balanji, Hamid Majidi

doi: 10.48550/arxiv.2110.03574pmid: N/A

Abstract:In this work, a simple vision algorithm is designed and implemented to extract and identify the surface defects on the Golden Delicious apples caused by the enzymic browning process. 34 Golden Delicious apples were selected for the experiments, of which 17 had enzymic browning defects and the other 17 were sound. The image processing part of the proposed vision algorithm extracted the defective surface area of the apples with high accuracy of 97.15%. The area and mean of the segmented images were selected as the 2x1 feature vectors to feed into a designed artificial neural network. The analysis based on the above features indicated that the images with a mean less than 0.0065 did not belong to the defective apples; rather, they were extracted as part of the calyx and stem of the healthy apples. The classification accuracy of the neural network applied in this study was 99.19%

Riess, Anneliese; Ziller, Alexander; Kolek, Stefan; Rueckert, Daniel; Schnabel, Julia; Kaissis, Georgios

doi: 10.48550/arxiv.2110.03478pmid: N/A

Abstract:Federated learning enhanced with Differential Privacy (DP) is a powerful privacy-preserving strategy to protect individuals sharing their sensitive data for processing in fields such as medicine and healthcare. Many medical applications, for example magnetic resonance imaging (MRI), rely on complex-valued signal processing techniques for data acquisition and analysis. However, the appropriate application of DP to complex-valued data is still underexplored. To address this issue, from the theoretical side, we introduce the complex-valued Gaussian mechanism, whose behaviour we characterise in terms of $f$-DP, $(\varepsilon, \delta)$-DP and Rényi-DP. Moreover, we generalise the fundamental algorithm DP stochastic gradient descent to complex-valued neural networks and present novel complex-valued neural network primitives compatible with DP. Experimentally, we showcase a proof-of-concept by training federated complex-valued neural networks with DP on a real-world task (MRI pulse sequence classification in $k$-space), yielding excellent utility and privacy. Our results highlight the relevance of combining federated learning with robust privacy-preserving techniques in the MRI context.

Chamakh, Linda; Szabo, Zoltan

doi: 10.48550/arxiv.2110.09516pmid: N/A

Abstract:Kernel techniques are among the most popular and flexible approaches in data science allowing to represent probability measures without loss of information under mild conditions. The resulting mapping called mean embedding gives rise to a divergence measure referred to as maximum mean discrepancy (MMD) with existing quadratic-time estimators (w.r.t. the sample size) and known convergence properties for bounded kernels. In this paper we focus on the problem of MMD estimation when the mean embedding of one of the underlying distributions is available analytically. Particularly, we consider distributions on the real line (motivated by financial applications) and prove tighter concentration for the proposed estimator under this semi-explicit setting; we also extend the result to the case of unbounded (exponential) kernel with minimax-optimal lower bounds. We demonstrate the efficiency of our approach beyond synthetic example in three real-world examples relying on one-dimensional random variables: index replication and calibration on loss-given-default ratios and on S&P 500 data.

Fröhlich, Lukas P.; Küttel, Christian; Arcari, Elena; Hewing, Lukas; Zeilinger, Melanie N.; Carron, Andrea

doi: 10.1109/iros47612.2022.9981780pmid: N/A

Abstract:Learning-based model predictive control has been widely applied in autonomous racing to improve the closed-loop behaviour of vehicles in a data-driven manner. When environmental conditions change, e.g., due to rain, often only the predictive model is adapted, but the controller parameters are kept constant. However, this can lead to suboptimal behaviour. In this paper, we address the problem of data-efficient controller tuning, adapting both the model and objective simultaneously. The key novelty of the proposed approach is that we leverage a learned dynamics model to encode the environmental condition as a so-called context. This insight allows us to employ contextual Bayesian optimization to efficiently transfer knowledge across different environmental conditions. Consequently, we require fewer data to find the optimal controller configuration for each context. The proposed framework is extensively evaluated with more than 3'000 laps driven on an experimental platform with 1:28 scale RC race cars. The results show that our approach successfully optimizes the lap time across different contexts requiring fewer data compared to other approaches based on standard Bayesian optimization.

Bruyère, Véronique; Pérez, Guillermo A.; Staquet, Gaëtan

doi: 10.1007/978-3-031-30823-9_14pmid: N/A

Abstract:We present a new learning algorithm for realtime one-counter automata. Our algorithm uses membership and equivalence queries as in Angluin's L* algorithm, as well as counter value queries and partial equivalence queries. In a partial equivalence query, we ask the teacher whether the language of a given finite-state automaton coincides with a counter-bounded subset of the target language. We evaluate an implementation of our algorithm on a number of random benchmarks and on a use case regarding efficient JSON-stream validation.