arXiv subjects

Kai Xiong

Publications and source records attributed to Kai Xiong.

At least 37 records · Page 2Linked to original sources

arXiv2025

UFO-RL: Uncertainty-Focused Optimization for Efficient Reinforcement Learning Data Selection

Scaling RL for LLMs is computationally expensive, largely due to multi-sampling for policy optimization and evaluation, making efficient data selection crucial. Inspired by the Zone of Proximal Development (ZPD) theory, we hypothesize LLMs learn best from data within their potential comprehension zone. Addressing the limitation of conventional, computationally intensive multi-sampling methods for data assessment, we introduce UFO-RL. This novel framework uses a computationally efficient single-pass uncertainty estimation to identify informative data instances, achieving up to 185x faster data evaluation. UFO-RL leverages this metric to select data within the estimated ZPD for training. Experiments show that training with just 10% of data selected by UFO-RL yields performance comparable to or surpassing full-data training, reducing overall training time by up to 16x while enhancing stability and generalization. UFO-RL offers a practical and highly efficient strategy for scaling RL fine-tuning of LLMs by focusing learning on valuable data.

cs.LG↗

arXiv2025

Beyond Similarity: A Gradient-based Graph Method for Instruction Tuning Data Selection

Large language models (LLMs) have shown great potential across various industries due to their remarkable ability to generalize through instruction tuning. However, the limited availability of domain-specific data significantly hampers their performance on specialized tasks. While existing methods primarily focus on selecting training data from general datasets that are similar to the target domain, they often fail to consider the joint distribution of instructions, resulting in inefficient learning and suboptimal knowledge transfer. To address these challenges, we introduce G2IS (Gradient-based Graph Instruction Selection), a novel method that constructs a mixed gradient-based instruction graph to capture the joint distribution and interdependencies between instructions. By accounting for the relationships between instructions, G2IS improves domain adaptation efficiency. Additionally, we propose a gradient walk algorithm to refine the data selection process, enhancing both training effectiveness and efficiency. Our experiments demonstrate that G2IS outperforms traditional methods across various domain adaptation tasks, yielding significant performance gains, particularly in complex, data-scarce scenarios. These results underscore the potential of G2IS in advancing the development of large, domain-specific models.

cs.CL↗

arXiv2025

Network Slice-based Low-Altitude Intelligent Network for Advanced Air Mobility

Advanced Air Mobility (AAM) is transforming transportation systems by extending them into near-ground airspace, offering innovative solutions to mobility challenges. In this space, electric vertical take-off and landing vehicles (eVTOLs) perform a variety of tasks to improve aviation safety and efficiency, such as collaborative computing and perception. However, eVTOLs face constraints such as compacted shape and restricted onboard computing resources. These limitations necessitate task offloading to nearby high-performance base stations (BSs) for timely processing. Unfortunately, the high mobility of eVTOLs, coupled with their restricted flight airlines and heterogeneous resource management creates significant challenges in dynamic task offloading. To address these issues, this paper introduces a novel network slice-based Low-Altitude Intelligent Network (LAIN) framework for eVTOL tasks. By leveraging advanced network slicing technologies from 5G/6G, the proposed framework dynamically adjusts communication bandwidth, beam alignment, and computing resources to meet fluctuating task demands. Specifically, the framework includes an access pairing method to pre-schedule optimal eVTOL-BS-slice assignments, a pre-assessment algorithm to avoid resource waste, and a deep reinforcement learning-based slice orchestration mechanism to optimize resource allocation and lifecycle management. Simulation results demonstrate that the proposed framework outperforms existing benchmarks in terms of resource allocation efficiency and operational/violation costs across varying eVTOL velocities. This work provides valuable insights into intelligent network slicing for future AAM transportation systems.

cs.NI↗

arXiv2025

Digital Twin-based SIM Communication and Flight Control for Advanced Air Mobility

Electric Vertical Take-off and Landing vehicles (eVTOLs) are driving Advanced Air Mobility (AAM) toward transforming urban transportation by extending travel from congested ground networks to low-altitude airspace. This transition promises to reduce traffic congestion and significantly shorten commute times. To ensure aviation safety, eVTOLs must fly within prescribed flight corridors. These corridors are managed by ground-based Air Traffic Control (ATCo) stations, which oversee air-ground communication and flight scheduling. However, one critical challenge remains: the lack of high rate air-ground communication and safe flight planning within these corridors. The introduction of 6G-oriented Stacked Intelligent Metasurface (SIM) technology presents a high rate communication solution. With advanced phase-shifting capabilities, SIM enables precise wireless signal control and supports beam-tracking communication with eVTOLs. Leveraging this technology, we propose a Composite Potential Field (CPF) approach. This method dynamically integrates target, separation, and communication fields to optimize both SIM communication efficiency and flight safety. Simulation results validate the effectiveness of this DT-based approach. Compared to the potential field flight control benchmark, it improves the transmission rate by 8.3\%. Additionally, it reduces flight distance deviation from the prescribed corridor by 10\% compared to predetermined optimization methods.

cs.NI↗

arXiv2024

Meaningful Learning: Enhancing Abstract Reasoning in Large Language Models via Generic Fact Guidance

Large language models (LLMs) have developed impressive performance and strong explainability across various reasoning scenarios, marking a significant stride towards mimicking human-like intelligence. Despite this, when tasked with several simple questions supported by a generic fact, LLMs often struggle to abstract and apply the generic fact to provide consistent and precise answers, revealing a deficiency in abstract reasoning abilities. This has sparked a vigorous debate about whether LLMs are genuinely reasoning or merely memorizing. In light of this, we design a preliminary study to quantify and delve into the abstract reasoning abilities of existing LLMs. Our findings reveal a substantial discrepancy between their general reasoning and abstract reasoning performances. To relieve this problem, we tailor an abstract reasoning dataset (AbsR) together with a meaningful learning paradigm to teach LLMs how to leverage generic facts for reasoning purposes. The results show that our approach not only boosts the general reasoning performance of LLMs but also makes considerable strides towards their capacity for abstract reasoning, moving beyond simple memorization or imitation to a more nuanced understanding and application of generic facts. The code is available at https://github.com/Waste-Wood/MeanLearn.

cs.CL↗

arXiv2024

Multi-hop RIS-aided Learning Model Sharing for Urban Air Mobility

Urban Air Mobility (UAM), powered by flying cars, is poised to revolutionize urban transportation by expanding vehicle travel from the ground to the air. This advancement promises to alleviate congestion and enable faster commutes. However, the fast travel speeds mean vehicles will encounter vastly different environments during a single journey. As a result, onboard learning systems need access to extensive environmental data, leading to high costs in data collection and training. These demands conflict with the limited in-vehicle computing and battery resources. Fortunately, learning model sharing offers a solution. Well-trained local Deep Learning (DL) models can be shared with other vehicles, reducing the need for redundant data collection and training. However, this sharing process relies heavily on efficient vehicular communications in UAM. To address these challenges, this paper leverages the multi-hop Reconfigurable Intelligent Surface (RIS) technology to improve DL model sharing between distant flying cars. We also employ knowledge distillation to reduce the size of the shared DL models and enable efficient integration of non-identical models at the receiver. Our approach enhances model sharing and onboard learning performance for cars entering new environments. Simulation results show that our scheme improves the total reward by 85% compared to benchmark methods.

cs.NI↗

arXiv2024

Supervised Fine-Tuning Achieve Rapid Task Adaption Via Alternating Attention Head Activation Patterns

LLMs' performance on complex tasks is still unsatisfactory. A key issue is that presently LLMs learn in a data-driven schema, while the instructions about these complex tasks are both scarce and hard to collect or construct. On the contrary, a prominent phenomenon is that LLMs can learn rather fast on simpler tasks with adequate prior knowledge captured during pretraining stage. Thus, if the prerequisite and mechanism of such rapid generalization could be elucidated, it could enhance the efficiency and effectiveness of the LLM's ability to learn complex tasks. Thus, in this paper, we employ a gradient-based method, to dissect the process that the SFT process adapts LLMs to downstream tasks via the perspective of attention patterns. We find that: (1) LLMs selectively activate task-specific attention heads during SFT; (2) activation patterns for complex tasks are combinations of basic task patterns; and (3) changes in a few parameters can significantly impact activation patterns after SFT on a small number of samples.Based on these insights, experiments are conducted to actually enhance the efficiency and effectiveness of SFT.

cs.LG↗

arXiv2024

RIS-aided Trajectory Optimization in Layered Urban Air Mobility

Urban Air Mobility (UAM) relies on developing aerospace industries, where safe aviation and efficient communication are critical features of aircraft. However, it is challenging for aircraft to sustain efficient air-ground communication in urban circumstances. Without continuous air-ground communication, aircraft may experience course deviation and safety accidents. To address these problems, a reconfigurable intelligent surface(RIS)-aided trajectory optimization scheme is proposed enabling efficient air-ground communication and safe aviation in UAM with a layered airspace structure. This paper first devises a dual-plane RIS communication scheme for layered airspace. It fully engages the omnidirectional and directional signal attributes to reduce the transmission delay of the air-ground communication. Based on the dual-plane RIS configuration, we jointly develop the intra- and inter-layer trajectory scheme to optimize communication and safe aviation. In the intra-layer trajectory optimization, we propose a dual-time-scale flight scheme to improve communication capacity and horizontal flight safety. Meanwhile, we propose a safe layer-switching method to ensure collision avoidance during vertical flight in the inter-layer trajectory optimization. The communication load of the proposed scheme can be improved 40% and the time of safe separation restoration can be lessened 66% compared with the benchmarks in the layered airspace.

cs.CE↗

arXiv2024

Deciphering the Impact of Pretraining Data on Large Language Models through Machine Unlearning

Through pretraining on a corpus with various sources, Large Language Models (LLMs) have gained impressive performance. However, the impact of each component of the pretraining corpus remains opaque. As a result, the organization of the pretraining corpus is still empirical and may deviate from the optimal. To address this issue, we systematically analyze the impact of 48 datasets from 5 major categories of pretraining data of LLMs and measure their impacts on LLMs using benchmarks about nine major categories of model capabilities. Our analyses provide empirical results about the contribution of multiple corpora on the performances of LLMs, along with their joint impact patterns, including complementary, orthogonal, and correlational relationships. We also identify a set of ``high-impact data'' such as Books that is significantly related to a set of model capabilities. These findings provide insights into the organization of data to support more efficient pretraining of LLMs.

cs.CL↗

arXiv2024

Diagnosing and Remedying Knowledge Deficiencies in LLMs via Label-free Curricular Meaningful Learning

Large Language Models (LLMs) are versatile and demonstrate impressive generalization ability by mining and learning information from extensive unlabeled text. However, they still exhibit reasoning mistakes, often stemming from knowledge deficiencies, which can affect their trustworthiness and reliability. Although users can provide diverse and comprehensive queries, obtaining sufficient and effective feedback is demanding. Furthermore, evaluating LLMs comprehensively with limited labeled samples is difficult. This makes it a challenge to diagnose and remedy the deficiencies of LLMs through rich label-free user queries. To tackle this challenge, we propose a label-free curricular meaningful learning framework (LaMer). LaMer first employs relative entropy to automatically diagnose and quantify the knowledge deficiencies of LLMs in a label-free setting. Next, to remedy the diagnosed knowledge deficiencies, we apply curricular meaningful learning: first, we adopt meaningful learning to adaptively synthesize augmentation data according to the severity of the deficiencies, and then design a curricular deficiency remedy strategy to remedy the knowledge deficiencies of LLMs progressively. Experiments show that LaMer efficiently and effectively diagnoses and remedies knowledge deficiencies in LLMs, improving various LLMs across seven out-of-distribution (OOD) reasoning and language understanding benchmarks, achieving comparable results to baselines with just 40\% training data. LaMer even surpasses methods that rely on labeled datasets for deficiency diagnosis. In application, our label-free method can offer an effective knowledge deficiency diagnostic tool for efficient LLM development.

cs.CL↗

arXiv2024

PACE: A Large-Scale Dataset with Pose Annotations in Cluttered Environments

We introduce PACE (Pose Annotations in Cluttered Environments), a large-scale benchmark designed to advance the development and evaluation of pose estimation methods in cluttered scenarios. PACE provides a large-scale real-world benchmark for both instance-level and category-level settings. The benchmark consists of 55K frames with 258K annotations across 300 videos, covering 238 objects from 43 categories and featuring a mix of rigid and articulated items in cluttered scenes. To annotate the real-world data efficiently, we develop an innovative annotation system with a calibrated 3-camera setup. Additionally, we offer PACE-Sim, which contains 100K photo-realistic simulated frames with 2.4M annotations across 931 objects. We test state-of-the-art algorithms in PACE along two tracks: pose estimation, and object pose tracking, revealing the benchmark's challenges and research opportunities. Our benchmark code and data is available on https://github.com/qq456cvb/PACE.

cs.CV↗

arXiv2024

RIS-empowered Topology Control for Distributed Learning in Urban Air Mobility

Urban Air Mobility (UAM) expands vehicles from the ground to the near-ground space, envisioned as a revolution for transportation systems. Comprehensive scene perception is the foundation for autonomous aerial driving. However, UAM encounters the intelligent perception challenge: high perception learning requirements conflict with the limited sensors and computing chips of flying cars. To overcome the challenge, federated learning (FL) and other collaborative learning have been proposed to enable resource-limited devices to conduct onboard deep learning (DL) collaboratively. But traditional collaborative learning like FL relies on a central integrator for DL model aggregation, which is difficult to deploy in dynamic environments. The fully decentralized learning schemes may be the intuitive solution while the convergence of distributed learning cannot be guaranteed. Accordingly, this paper explores reconfigurable intelligent surfaces (RIS) empowered distributed learning, taking account of topological attributes to facilitate the learning performance with convergence guarantee. We propose several FL topological criteria for optimizing the transmission delay and convergence rate by exploiting the Laplacian matrix eigenvalues of the communication network. Subsequently, we innovatively leverage the RIS link modification ability to remold the current network according to the proposed topological criteria. This paper rethinks the functions of RIS from the perspective of the network layer. Furthermore, a deep deterministic policy gradient-based RIS phase shift control algorithm is developed to construct or deconstruct the network links simultaneously to reshape the communication network. Simulation experiments are conducted over MobileNet-based multi-view learning to verify the efficiency of the distributed FL framework.

cs.IT↗

arXiv2024

Intuitive or Dependent? Investigating LLMs' Behavior Style to Conflicting Prompts

This study investigates the behaviors of Large Language Models (LLMs) when faced with conflicting prompts versus their internal memory. This will not only help to understand LLMs' decision mechanism but also benefit real-world applications, such as retrieval-augmented generation (RAG). Drawing on cognitive theory, we target the first scenario of decision-making styles where there is no superiority in the conflict and categorize LLMs' preference into dependent, intuitive, and rational/irrational styles. Another scenario of factual robustness considers the correctness of prompt and memory in knowledge-intensive tasks, which can also distinguish if LLMs behave rationally or irrationally in the first scenario. To quantify them, we establish a complete benchmarking framework including a dataset, a robustness evaluation pipeline, and corresponding metrics. Extensive experiments with seven LLMs reveal their varying behaviors. And, with role play intervention, we can change the styles, but different models present distinct adaptivity and upper-bound. One of our key takeaways is to optimize models or the prompts according to the identified style. For instance, RAG models with high role play adaptability may dynamically adjust the interventions according to the quality of retrieval results -- being dependent to better leverage informative context; and, being intuitive when external prompt is noisy.

cs.CL↗

arXiv2023

Examining Inter-Consistency of Large Language Models Collaboration: An In-depth Analysis via Debate

Large Language Models (LLMs) have shown impressive capabilities in various applications, but they still face various inconsistency issues. Existing works primarily focus on the inconsistency issues within a single LLM, while we complementarily explore the inter-consistency among multiple LLMs for collaboration. To examine whether LLMs can collaborate effectively to achieve a consensus for a shared goal, we focus on commonsense reasoning, and introduce a formal debate framework (FORD) to conduct a three-stage debate among LLMs with real-world scenarios alignment: fair debate, mismatched debate, and roundtable debate. Through extensive experiments on various datasets, LLMs can effectively collaborate to reach a consensus despite noticeable inter-inconsistencies, but imbalances in their abilities can lead to domination by superior LLMs. Leveraging a more advanced LLM like GPT-4 as an authoritative judge can boost collaboration performance. Our work contributes to understanding the inter-consistency among LLMs and lays the foundation for developing future collaboration methods. Codes and data are available at https://github.com/Waste-Wood/FORD

cs.CL↗

arXiv2023

Improving Cross-Task Generalization with Step-by-Step Instructions

Instruction tuning has been shown to be able to improve cross-task generalization of language models. However, it is still challenging for language models to complete the target tasks following the instructions, as the instructions are general and lack intermediate steps. To address this problem, we propose to incorporate the step-by-step instructions to help language models to decompose the tasks, which can provide the detailed and specific procedures for completing the target tasks. The step-by-step instructions are obtained automatically by prompting ChatGPT, which are further combined with the original instructions to tune language models. The extensive experiments on SUP-NATINST show that the high-quality step-by-step instructions can improve cross-task generalization across different model sizes. Moreover, the further analysis indicates the importance of the order of steps of the step-by-step instruction for the improvement. To facilitate future research, we release the step-by-step instructions and their human quality evaluation results.

cs.CL↗

arXiv2023

A Digital Twin Empowered Lightweight Model Sharing Scheme for Multi-Robot Systems

Multi-robot system for manufacturing is an Industry Internet of Things (IIoT) paradigm with significant operational cost savings and productivity improvement, where Unmanned Aerial Vehicles (UAVs) are employed to control and implement collaborative productions without human intervention. This mission-critical system relies on 3-Dimension (3-D) scene recognition to improve operation accuracy in the production line and autonomous piloting. However, implementing 3-D point cloud learning, such as Pointnet, is challenging due to limited sensing and computing resources equipped with UAVs. Therefore, we propose a Digital Twin (DT) empowered Knowledge Distillation (KD) method to generate several lightweight learning models and select the optimal model to deploy on UAVs. With a digital replica of the UAVs preserved at the edge server, the DT system controls the model sharing network topology and learning model structure to improve recognition accuracy further. Moreover, we employ network calculus to formulate and solve the model sharing configuration problem toward minimal resource consumption, as well as convergence. Simulation experiments are conducted over a popular point cloud dataset to evaluate the proposed scheme. Experiment results show that the proposed model sharing scheme outperforms the individual model in terms of computing resource consumption and recognition accuracy.

cs.IT↗

arXiv2022

ReCo: Reliable Causal Chain Reasoning via Structural Causal Recurrent Neural Networks

Causal chain reasoning (CCR) is an essential ability for many decision-making AI systems, which requires the model to build reliable causal chains by connecting causal pairs. However, CCR suffers from two main transitive problems: threshold effect and scene drift. In other words, the causal pairs to be spliced may have a conflicting threshold boundary or scenario. To address these issues, we propose a novel Reliable Causal chain reasoning framework~(ReCo), which introduces exogenous variables to represent the threshold and scene factors of each causal pair within the causal chain, and estimates the threshold and scene contradictions across exogenous variables via structural causal recurrent neural networks~(SRNN). Experiments show that ReCo outperforms a series of strong baselines on both Chinese and English CCR datasets. Moreover, by injecting reliable causal chain knowledge distilled by ReCo, BERT can achieve better performances on four downstream causal-related tasks than BERT models enhanced by other kinds of knowledge.

cs.AI↗

arXiv2022

One-Shot General Object Localization

This paper presents a general one-shot object localization algorithm called OneLoc. Current one-shot object localization or detection methods either rely on a slow exhaustive feature matching process or lack the ability to generalize to novel objects. In contrast, our proposed OneLoc algorithm efficiently finds the object center and bounding box size by a special voting scheme. To keep our method scale-invariant, only unit center offset directions and relative sizes are estimated. A novel dense equalized voting module is proposed to better locate small texture-less objects. Experiments show that the proposed method achieves state-of-the-art overall performance on two datasets: OnePose dataset and LINEMOD dataset. In addition, our method can also achieve one-shot multi-instance detection and non-rigid object localization. Code repository: https://github.com/qq456cvb/OneLoc.

cs.CV↗