Yanyan Shen

The Shapley value provides a theoretically grounded framework for model interpretability by quantifying feature importance, with critical applications in data analytics and data valuation. However, its exact computation is often intractable due to an exponential complexity that scales with the number of features. This poses a significant scalability challenge for high-dimensional datasets common in modern data systems. Prevailing approximation methods, typically based on Monte Carlo sampling, suffer from severe inefficiencies rooted in their unstructured and redundant exploration of the coalition space. These limitations lead to unreliable estimates and hinder practical adoption in data-intensive applications. This paper presents CoShap, a novel framework that delivers fast and reliable Shapley value approximation through systematic coalition growth. CoShap consists of a two-phase evaluation: (1) A layer-wise coalition growth phase that incrementally builds larger feature coalitions from smaller ones. Inspired by Pattern Growth, this structured approach drastically reduces redundant model evaluations by reusing previously computed coalition utilities. (2) A subsequent feature-wise evaluation phase that adaptively allocates more computational budget to high-variance features, specifically those with the most uncertain contributions, to enhance approximation reliability. To guarantee an (ε, δ)-approximation of the true Shapley values, CoShap introduces a dynamic budget allocation method that balances the evaluation budget across both phases. Experiments on diverse datasets show that CoShap significantly outperforms state-of-the-art methods, achieving up to 7.15× speedup in runtime and a 76.84% reduction in approximation error.

Predicting stock prices presents a challenging research problem due to the inherent volatility and non-linear nature of the stock market. In recent years, knowledge-enhanced stock price prediction methods have shown groundbreaking results by utilizing external knowledge to understand the stock market. Despite the importance of these methods, there is a scarcity of scholarly works that systematically synthesize previous studies from the perspective of external knowledge types. Specifically, the external knowledge can be modeled in different data structures, which we group into non-graph-based formats and graph-based formats: (1) non-graph-based knowledge captures contextual information and multimedia descriptions specifically associated with an individual stock; (2) graph-based knowledge captures interconnected and interdependent information in the stock market. This survey article aims to provide a systematic and comprehensive description of methods for acquiring external knowledge from various unstructured data sources and then incorporating it into stock price prediction models. We also explore fusion methods for combining external knowledge with historical price features. Moreover, this article includes a compilation of relevant datasets and delves into potential future research directions in this domain.

Large language model (LLM) inference serving systems are essential to various LLM-based applications. As demand for LLM services continues to grow, scaling these systems to handle high request rates while meeting latency Service-Level Objectives (SLOs), referred to as effective throughput, becomes critical. However, existing systems often struggle to improve effective throughput, primarily due to a significant decline in Time To First Token (TTFT) SLO attainment. We identify two major causes of this bottleneck: (1) memory-intensive KV cache that limits batch size expansion under GPU memory constraints, and (2) rigid batch composition enforced by the default First-Come-First-Serve scheduling policy. In this paper, we introduce Apt-Serve, a scalable framework designed to enhance effective throughput in LLM inference serving. Apt-Serve features a new hybrid cache scheme that combines KV cache with a memory-efficient hidden cache for reusable input hidden state vectors, allowing large batch sizes and improving request concurrency. Based on the hybrid cache, Apt-Serve employs an adaptive runtime scheduling mechanism that dynamically optimizes batch composition. We formally define the adaptive scheduling optimization problem and propose an efficient algorithm with theoretical guarantees. Extensive evaluations on three real-world datasets and LLMs ranging from 13B to 66B parameters demonstrate that Apt-Serve achieves up to 8.8x improvement in effective throughput compared to the state-of-the-art inference serving systems.

As large language models (LLMs) have shown great success in many tasks, they are used in various applications. While a lot of works have focused on the efficiency of single-LLM application (e.g., offloading, request scheduling, parallelism strategy selection), multi-LLM applications receive less attention, particularly in offline inference scenarios. In this work, we aim to improve the offline end-to-end inference efficiency of multi-LLM applications in the single-node multi-GPU environment. The problem involves two key decisions: (1) determining which LLMs to run concurrently each time (we may not run all the models at the same time), and (2) selecting a parallelism strategy to use for each LLM. This problem is NP-hard. Naive solutions may not work well because the running time for a model to complete a set of requests depends on the request workload and the selected parallelism strategy, and they lack an accurate model of the running time. As the LLM output lengths are unknown before running, to estimate the model running time, we propose a sampling-then-simulation method which first estimates the output lengths by sampling from an empirical cumulative function we obtained from a large dataset in advance, and then simulates the LLM inference process accordingly. Based on the simulation, we estimate the per-iteration latencys to get the total latency. A greedy method is proposed to optimize the scheduling of the LLMs in the application across the GPUs. We then propose a framework SamuLLM which contains two phases: planning, which calls the greedy method for an application and running, which runs the application and dynamically adjust the model scheduling based on the runtime information. Experiments on 3 applications and a mixed application show that SamuLLM can achieve 1.0-2.4$\times$ end-to-end speedups compared to the competitors.