This work aims at improving the energy efficiency of decentralized learning by optimizing the mixing matrix, which controls the communication demands during the learning process. Through rigorous analysis based on a state-of-the-art decentralized learning algorithm, the problem is formulated as a bi-level optimization, with the lower level solved by graph sparsification. A solution with guaranteed performance is proposed for the special case of fully-connected base topology and a greedy heuristic is proposed for the general case. Simulations based on real topology and dataset show that the proposed solution can lower the energy consumption at the busiest node by 54%–76% while maintaining the quality of the trained model.

We consider a novel active learning problem motivated by the need of learning machine learning models for health monitoring in wireless body area network (WBAN). Due to the limited resources at body sensors, collecting each unlabeled sample in WBAN incurs a nontrivial cost. Moreover, training health monitoring models typically requires labels indicating the patient's health state that need to be generated by healthcare professionals, which cannot be obtained at the same pace as data collection. These challenges make our problem fundamentally different from classical active learning, where unlabeled samples are free and labels can be queried in real time. To handle these challenges, we propose a two-phased active learning method, consisting of an online phase where a coreset construction algorithm is proposed to select a subset of unlabeled samples based on their noisy predictions, and an offline phase where the selected samples are labeled to train the target model. The samples selected by our algorithm are proved to yield a guaranteed error in approximating the full dataset in evaluating the loss function. Our evaluation based on real health monitoring data and our own experimentation demonstrates that our solution can drastically save the data curation cost without sacrificing the quality of the target model.

We consider the problem of training a given machine learning model by decentralized parallel stochastic gradient descent over training data distributed across multiple nodes, which arises in many application scenarios. Although extensive studies have been conducted on improving the communication efficiency by optimizing what to communicate between nodes (e.g., model compression) and how often to communicate, recent studies have shown that it is also important to customize the communication patterns between each pair of nodes, which is the focus of this work. To this end, we propose a framework and efficient algorithms to design the communication patterns through Laplacian matrix sampling (LMS), which governs not only which nodes should communicate with each other but also what weights the communicated parameters should carry during parameter aggregation. Our framework is designed to minimize the total cost incurred until convergence based on any given cost model that is additive over iterations, with focus on minimizing the communication cost. Besides achieving a theoretically guaranteed performance in the special case of additive homogeneous communication costs, our solution also achieves superior performance under a variety of network settings and cost models in experiments based on real datasets and topologies, saving 24–50% of the cost compared to the state-of-the-art design without compromising the quality of the trained model.

Network tomography is a powerful tool to monitor the internal state of a closed network that cannot be measured directly, with broad applications in the Internet, overlay networks, and all-optical networks. However, existing network tomography solutions all assume that the measurements are trust-worthy, leaving open how effective they are in an adversarial environment with possibly manipulated measurements. To understand the fundamental limit of network tomography in such a setting, we formulate and analyze a novel type of attack that aims at maximally degrading the performance of targeted paths without being localized by network tomography. By analyzing properties of the optimal attack strategy, we formulate novel combinatorial optimizations to design the optimal attack strategy, which are then linked to well-known NP-hard problems and approximation algorithms. As a byproduct, our algorithms also identify approximations of the most vulnerable set of links that once manipulated, can inflict the maximum performance degradation. Our evaluations on real topologies demonstrate the large potential damage of such attacks, signaling the need of new defenses.

As a tool to infer the internal state of a network that cannot be measured directly (e.g., the Internet and all-optical networks), network tomography has been extensively studied under the assumption that the measurements truthfully reflect the end-to-end performance of measurement paths, which makes the resulting solutions vulnerable to manipulated measurements. In this work, we investigate the impact of manipulated measurements via a recently proposed attack model called the stealthy DeGrading of Service (DGoS) attack, which aims at maximally degrading path performances without exposing the manipulated links to network tomography. While existing studies on this attack assume that network tomography only measures the paths actively used for data transfer (by passively recording the performance of data packets), our model allows network tomography to measure a larger set of paths, e.g., by sending probes on some paths not carrying data flows. By developing and analyzing the optimal attack strategy, we quantify the maximum damage of such an attack and shed light on possible defenses.

Network tomography is a powerful tool to monitor the internal state of a closed network that cannot be measured directly, with broad applications in the Internet, overlay networks, and all-optical networks. However, existing network tomography solutions all assume that the measurements are trust-worthy, leaving open how effective they are in an adversarial environment with possibly manipulated measurements. To understand the fundamental limit of network tomography in such a setting, we formulate and analyze a novel type of attack that aims at maximally degrading the performance of targeted paths without being localized by network tomography. By analyzing properties of the optimal attack, we formulate novel combinatorial optimizations to design the optimal attack strategy, which are then linked to well-known problems and approximation algorithms. Our evaluations on real topologies demonstrate the large damage of such attacks, signaling the need of new defenses.

We consider the problem of privately releasing a class of queries that we call hierarchical count-of-counts histograms . Count-of-counts histograms partition the rows of an input table into groups (e.g., group of people in the same household), and for every integer j report the number of groups of size j . Hierarchical count-of-counts queries report count-of-counts histograms at different granularities as per hierarchy defined on an attribute in the input data (e.g., geographical location of a household at the national, state and county levels). In this paper, we introduce this problem, along with appropriate error metrics and propose a differentially private solution that generates count-of-counts histograms that are consistent across all levels of the hierarchy.