Graph Neural Networks (GNNs) have emerged as powerful tools for supervised machine learning over graph-structured data, while sampling-based node representation learning is widely utilized in unsupervised learning. However, scalability remains a major challenge in both supervised and unsupervised learning for large graphs (e.g., those with over 1 billion nodes). The scalability bottleneck largely stems from the mini-batch sampling phase in GNNs and the random walk sampling phase in unsupervised methods. These processes often require storing features or embeddings in memory. In the context of distributed training, they require frequent, inefficient random access to data stored across different workers. Such repeated inter-worker communication for each mini-batch leads to high communication overhead and computational inefficiency.

As memory-intensive applications continue to drive the need for advanced architectural solutions, Compute Express Link (CXL) has risen as a promising interconnect technology that enables seamless high-speed, low-latency communication between host processors and various peripheral devices. In this study, we explore the application performance of ASIC CXL memory in various data-center scenarios. We then further explore multiple potential impacts (e.g., throughput, latency, and cost reduction) of employing CXL memory via carefully designed policies and strategies. Our empirical results show the high potential of CXL memory, reveal multiple intriguing observations of CXL memory and contribute to the wide adoption of CXL memory in real-world deployment environments. Based on our benchmarks, we also develop an Abstract Cost Model that can estimate the cost benefit from using CXL memory.

Web service composition (WSC) is the task of combining a chain of connected single services together to create a more complex and value-added composite service. Quality of service (QoS) has been mostly applied to represent nonfunctional properties of web services and differentiate those with the same functionality. Many research has been done on QoS-aware service composition, as it significantly affects the quality of a composite service. However, existing methods are restricted to predefined workflows, which can incur a couple of limitations, including the lack of guarantee for the optimality on overall QoS and for the completeness of finding a composite service solution. In this paper, instead of predefining a workflow model for service composition, we propose a novel planning-based approach that can automatically convert a QoS-aware composition task to a planning problem with temporal and numerical features. Furthermore, we use state-of-the-art planners, including an existing one and a self-developed one, to handle complex temporal planning problems with logical reasoning and numerical optimization. Our approach can find a composite service graph with the optimal overall QoS value while satisfying multiple global QoS constraints. We implement a prototype system and conduct extensive experiments on large web service repositories. The experimental results show that our proposed approach largely outperforms existing ones in terms of solution quality and is efficient enough for practical deployment.

Plan recognition is a ubiquitous task in artificial intelligence and pervasive computing research. The multigoal recognition problem presents a major challenge in the real world of plan recognition. Users often pursue several goals in a concurrent and interleaving manner, where the pursuit of goals may spread over different parts of an activity sequence and may be pursued in parallel. Existing approaches for multigoal problems are probabilistic approaches. They all assume the existence of plan libraries, which require a lot of human efforts in predicting and formalizing plans and may be impractical in many cases. In this paper, we present a novel logic-based approach to solve the multigoal recognition problem efficiently, without the need of plan libraries, using a state-of-the-art heuristic search planner LAMA. In particular, we first propose the formulation of a multigoal recognition problem based on automated planning. Then we present a bilevel probabilistic plan recognition approach that deals with both concurrent and interleaving goals from observed activity sequences. Experimental results over several domains show that our method has great flexibility and scalability.

Complex features, such as temporal dependencies and numerical cost constraints, are hallmarks of real-world planning problems. In this article, we consider the challenging problem of cost-sensitive temporally expressive (CSTE) planning, which requires concurrency of durative actions and optimization of action costs. We first propose a scheme to translate a CSTE planning problem to a minimum cost (MinCost) satisfiability (SAT) problem and to integrate with a relaxed parallel planning semantics for handling true temporal expressiveness. Our scheme finds solution plans that optimize temporal makespan, and also minimize total action costs at the optimal makespan. We propose two approaches for solving MinCost SAT. The first is based on a transformation of a MinCost SAT problem to a weighted partial Max-SAT (WPMax-SAT), and the second, called BB-CDCL, is an integration of the branch-and-bound technique and the conflict driven clause learning (CDCL) method. We also develop a CSTE customized variable branching scheme for BB-CDCL which can significantly improve the search efficiency. Our experiments on the existing CSTE benchmark domains show that our planner compares favorably to the state-of-the-art temporally expressive planners in both efficiency and quality.

For Web service composition, choreography has recently received great attention and demonstrated a few key advantages over orchestration such as distributed control, fairness, data efficiency, and scalability. Automated design of choreography plans, especially distributed plans for multiple roles, is more complex and has not been studied before. Existing work requires manual generation assisted by model checking. In this paper, we propose a novel planning-based approach that can automatically convert a given composition task to a distributed choreography specification. Although planning has been used for orchestration, it is difficult to use planning for choreography, as it involves decentralized control, concurrent workflows, and contingency. We propose a few novel techniques, including compilation of contingencies, dependency graph analysis, and communication control, to handle these characteristics using planning. We theoretically show the correctness of this approach and empirically evaluate its practicability.

Planning as satisfiability is a principal approach to planning with many eminent advantages. The existing planning as satisfiability techniques usually use encodings compiled from STRIPS. We introduce a novel SAT encoding scheme (SASE) based on the SAS+ formalism. The new scheme exploits the structural information in SAS+, resulting in an encoding that is both more compact and efficient for planning. We prove the correctness of the new encoding by establishing an isomorphism between the solution plans of SASE and that of STRIPS based encodings. We further analyze the transition variables newly introduced in SASE to explain why it accommodates modern SAT solving algorithms and improves performance. We give empirical statistical results to support our analysis. We also develop a number of techniques to further reduce the encoding size of SASE, and conduct experimental studies to show the strength of each individual technique. Finally, we report extensive experimental results to demonstrate significant improvements of SASE over the state-of-the-art STRIPS based encoding schemes in terms of both time and memory efficiency.

Cloud computing is emerging as a prominent computing model. It provides a low-cost, highly accessible alternative to other traditional high-performance computing platforms. It also has many other benefits such as high availability, scalability, elasticity, and free of maintenance. Given these attractive features, it is very desirable if automated planning can exploit the large, affordable computational power of cloud computing. However, the latency in inter-process communication in cloud computing makes most existing parallel planning algorithms unsuitable for cloud computing. In this paper, we propose a portfolio stochastic search framework that takes advantage of and is suitable for cloud computing. We first study the running time distribution of Monte-Carlo Random Walk (MRW) search, a stochastic planning algorithm, and show that the running time distribution usually has remarkable variability. Then, we propose a portfolio search algorithm that is suitable for cloud computing, which typically has abundant computing cores but high communication latency between cores. Further, we introduce an enhanced portfolio with multiple parameter settings to improve the efficiency of the algorithm. We implement the portfolio search algorithm in both a local cloud and the Windows Azure cloud. Experimental results show that our algorithm achieves good, in many cases super linear, speedup in the cloud platforms. Moreover, our algorithm greatly reduces the running time variance of the stochastic search and improves the solution quality. We also show that our scheme is economically sensible and robust under processor failures.

Planning as satisfiability is a principal approach to planning with many eminent advantages. The existing planning as satisfiability techniques usually use encodings compiled from the STRIPS formalism. We introduce a novel SAT encoding scheme based on the SAS+ formalism. It exploits the structural information in the SAS+ formalism, resulting in more compact SAT instances and reducing the number of clauses by up to 50 fold. Our results show that this encoding scheme improves upon the STRIPS-based encoding, in terms of both time and memory efficiency.