As the demand for cloud services and data-intensive applications continues to surge, the design of efficient and reliable data center network (DCN) topologies has become increasingly critical. However, traditional DCNs often face challenges of limited scalability, insufficient fault tolerance, and high communication latency. To address these issues, we introduce SFDC, a novel recursive and modular server-centric network topology. SFDC is built on a hierarchical element-layer structure that enables the construction of highly scalable and fault-tolerant networks. The modular design of SFDC supports flexible expansion, allowing for the integration of servers with varying network interface card (NIC) configurations without requiring significant redesigns. Furthermore, SFDC’s design effectively mitigates the growth of network diameter, ensuring low latency even at massive scales. We also propose a routing algorithm, SFRouting, which leverages SFDC’s hierarchical structure to efficiently compute unicast paths, while minimizing routing complexity and enhancing data transmission efficiency. Additionally, we present a multipath routing scheme based on disjoint path construction, which ensures robust communication by providing alternative paths in case of node or link failures, thus enhancing network fault tolerance. Experimental results demonstrate that SFDC outperforms existing DCN topologies such as BCube, DCell, and HS-DCell, exhibiting superior scalability, reduced network diameter, and enhanced fault tolerance while maintaining low latency and stable performance.

Nowadays the robustness of Deep Neural Networks (DNN) is gaining much more attention than ever. That is because DNNs are intensively adopted in safety-critical AI-enabled applications such as autonomous driving and authentication control. Formal methods have been proved to be effective to provide provable guarantee to the robustness of DNNs. However, they are suffering from bad scalability due to intrinsic high computational complexity of the verification problem. In this paper, we propose a novel attack-guided approach for efficiently verifying the robustness of neural networks. The novelty of our approach is that we use existing attack approaches to generate coarse adversarial examples, by which we can significantly simply final verification problem. In particular, we are focused on the neural networks that take ReLU activation functions, which are widely adopted for solving classification problems. The experimental results show that our approach outperforms those verification tools based on constraint solving by up to 69 times speedup, while it can compute minimum adversarial examples. The improvement is particularly significant on those adversarially trained networks.

Formal robustness verification of deep neural networks (DNNs) is a promising approach for achieving a provable reliability guarantee to AI-enabled software systems. Limited scalability is one of the main obstacles to the verification problem. In this paper, we propose eager falsification to accelerate the robustness verification of DNNs. It divides the verification problem into a set of independent subproblems and solves them in descending order of their falsification probabilities. Once a subproblem is falsified, the verification terminates with a conclusion that the network is not robust. We introduce a notion of label affinity to measure the falsification probability and present an approach to computing the probability based on symbolic interval propagation. Our approach is orthogonal to existing verification techniques. We integrate it into four state-of-the-art verification tools, i.e., MIPVerify, Neurify, DeepZ, and DeepPoly, and conduct extensive experiments on 8 benchmark datasets. The experimental results show that our approach can significantly improve these tools by up to 200x speedup when the perturbation distance is in a reasonable range.