Tianchi Huang

Lianchen Jia, Chaoyang Li, Ziqi Yuan et al.

Over the past decade, adaptive video streaming technology has witnessed significant advancements, particularly driven by the rapid evolution of deep learning techniques. However, the black-box nature of deep learning algorithms presents challenges for developers in understanding decision-making processes and optimizing for specific application scenarios. Although existing research has enhanced algorithm interpretability through decision tree conversion, interpretability does not directly equate to developers' subjective comprehensibility. To address this challenge, we introduce \texttt{ComTree}, the first bitrate adaptation algorithm generation framework that considers comprehensibility. The framework initially generates the complete set of decision trees that meet performance requirements, then leverages large language models to evaluate these trees for developer comprehensibility, ultimately selecting solutions that best facilitate human understanding and enhancement. Experimental results demonstrate that \texttt{ComTree} significantly improves comprehensibility while maintaining competitive performance, showing potential for further advancement. The source code is available at https://github.com/thu-media/ComTree.

Lianchen Jia, Chaoyang Li, Qian Houde et al.

Control algorithms in production environments typically require domain experts to tune their parameters and logic for specific scenarios. However, existing research predominantly focuses on algorithmic performance under ideal or default configurations, overlooking the critical aspect of Tuning Potential. To bridge this gap, we introduce Crucible, an agent that employs an LLM-driven, multi-level expert simulation to turn algorithms and defines a formalized metric to quantitatively evaluate their Tuning Potential. We demonstrate Crucible's effectiveness across a wide spectrum of case studies, from classic control tasks to complex computer systems, and validate its findings in a real-world deployment. Our experimental results reveal that Crucible systematically quantifies the tunable space across different algorithms. Furthermore, Crucible provides a new dimension for algorithm analysis and design, which ultimately leads to performance improvements. Our code is available at https://github.com/thu-media/Crucible.

Xin Yao, Tianchi Huang, Rui-Xiao Zhang et al.

Federated learning (FL) aims to train machine learning models in the decentralized system consisting of an enormous amount of smart edge devices. Federated averaging (FedAvg), the fundamental algorithm in FL settings, proposes on-device training and model aggregation to avoid the potential heavy communication costs and privacy concerns brought by transmitting raw data. However, through theoretical analysis we argue that 1) the multiple steps of local updating will result in gradient biases and 2) there is an inconsistency between the expected target distribution and the optimization objectives following the training paradigm in FedAvg. To tackle these problems, we first propose an unbiased gradient aggregation algorithm with the keep-trace gradient descent and the gradient evaluation strategy. Then we introduce an additional controllable meta updating procedure with a small set of data samples, indicating the expected target distribution, to provide a clear and consistent optimization objective. Both the two improvements are model- and task-agnostic and can be applied individually or together. Experimental results demonstrate that the proposed methods are faster in convergence and achieve higher accuracy with different network architectures in various FL settings.

Xin Yao, Tianchi Huang, Chenglei Wu et al.

Federated learning (FL) enables on-device training over distributed networks consisting of a massive amount of modern smart devices, such as smartphones and IoT (Internet of Things) devices. However, the leading optimization algorithm in such settings, i.e., federated averaging (FedAvg), suffers from heavy communication costs and the inevitable performance drop, especially when the local data is distributed in a non-IID way. To alleviate this problem, we propose two potential solutions by introducing additional mechanisms to the on-device training. The first (FedMMD) is adopting a two-stream model with the MMD (Maximum Mean Discrepancy) constraint instead of a single model in vanilla FedAvg to be trained on devices. Experiments show that the proposed method outperforms baselines, especially in non-IID FL settings, with a reduction of more than 20% in required communication rounds. The second is FL with feature fusion (FedFusion). By aggregating the features from both the local and global models, we achieve higher accuracy at fewer communication costs. Furthermore, the feature fusion modules offer better initialization for newly incoming clients and thus speed up the process of convergence. Experiments in popular FL scenarios show that our FedFusion outperforms baselines in both accuracy and generalization ability while reducing the number of required communication rounds by more than 60%.

Rui-Xiao Zhang, Tianchi Huang, Chenglei Wu et al.

Video caching has been a basic network functionality in today's network architectures. Although the abundance of caching replacement algorithms has been proposed recently, these methods all suffer from a key limitation: due to their immature rules, inaccurate feature engineering or unresponsive model update, they cannot strike a balance between the long-term history and short-term sudden events. To address this concern, we propose LA-E2, a long-short-term fusion caching replacement approach, which is based on a learning-aided exploration-exploitation process. Specifically, by effectively combining the deep neural network (DNN) based prediction with the online exploitation-exploration process through a \emph{top-k} method, LA-E2 can both make use of the historical information and adapt to the constantly changing popularity responsively. Through the extensive experiments in two real-world datasets, we show that LA-E2 can achieve state-of-the-art performance and generalize well. Especially when the cache size is small, our approach can outperform the baselines by 17.5\%-68.7\% higher in total hit rate.

Tianchi Huang, Xin Yao, Chenglei Wu et al.

Existing reinforcement learning~(RL)-based adaptive bitrate~(ABR) approaches outperform the previous fixed control rules based methods by improving the Quality of Experience~(QoE) score, as the QoE metric can hardly provide clear guidance for optimization, finally resulting in the unexpected strategies. In this paper, we propose \emph{Tiyuntsong}, a self-play reinforcement learning approach with generative adversarial network~(GAN)-based method for ABR video streaming. Tiyuntsong learns strategies automatically by training two agents who are competing against each other. Note that the competition results are determined by a set of rules rather than a numerical QoE score that allows clearer optimization objectives. Meanwhile, we propose GAN Enhancement Module to extract hidden features from the past status for preserving the information without the limitations of sequence lengths. Using testbed experiments, we show that the utilization of GAN significantly improves the Tiyuntsong's performance. By comparing the performance of ABRs, we observe that Tiyuntsong also betters existing ABR algorithms in the underlying metrics.

Tianchi Huang, Rui-Xiao Zhang, Chao Zhou et al.

Rate control is widely adopted during video streaming to provide both high video qualities and low latency under various network conditions. However, despite that many work have been proposed, they fail to tackle one major problem: previous methods determine a future transmission rate as a single for value which will be used in an entire time-slot, while real-world network conditions, unlike lab setup, often suffer from rapid and stochastic changes, resulting in the failures of predictions. In this paper, we propose a delay-constrained rate control approach based on end-to-end deep learning. The proposed model predicts future bit rate not as a single value, but as possible bit rate ranges using target delay gradient, with which the transmission delay is guaranteed. We collect a large scale of real-world live streaming data to train our model, and as a result, it automatically learns the correlation between throughput and target delay gradient. We build a testbed to evaluate our approach. Compared with the state-of-the-art methods, our approach demonstrates a better performance in bandwidth utilization. In all considered scenarios, a range based rate control approach outperforms the one without range by 19% to 35% in average QoE improvement.