Search

This paper studies the wireless scheduling design to coordinate the transmissions of (local) model parameters of federated learning (FL) for a swarm of unmanned aerial vehicles (UAVs). The overall goal of the proposed design is to realize the FL training and aggregation processes with a central aggregator exploiting the sensory data collected by the UAVs but it considers the multi-hop wireless network formed by the UAVs. Such transmissions of model parameters over the UAV-based wireless network potentially cause large transmission delays and overhead. Our proposed framework smartly aggregates local model parameters trained by the UAVs while efficiently transmitting the underlying parameters to the central aggregator in each FL global round. We theoretically show that the proposed scheme achieves minimal delay and communication overhead. Extensive numerical experiments demonstrate the superiority of the proposed scheme compared to other baselines., Comment: accepted to WCNC'24

Privacy policies have emerged as the predominant approach to conveying privacy notices to mobile application users. In an effort to enhance both readability and user engagement, the concept of contextual privacy policies (CPPs) has been proposed by researchers. The aim of CPPs is to fragment privacy policies into concise snippets, displaying them only within the corresponding contexts within the application's graphical user interfaces (GUIs). In this paper, we first formulate CPP in mobile application scenario, and then present a novel multimodal framework, named SeePrivacy, specifically designed to automatically generate CPPs for mobile applications. This method uniquely integrates vision-based GUI understanding with privacy policy analysis, achieving 0.88 precision and 0.90 recall to detect contexts, as well as 0.98 precision and 0.96 recall in extracting corresponding policy segments. A human evaluation shows that 77% of the extracted privacy policy segments were perceived as well-aligned with the detected contexts. These findings suggest that SeePrivacy could serve as a significant tool for bolstering user interaction with, and understanding of, privacy policies. Furthermore, our solution has the potential to make privacy notices more accessible and inclusive, thus appealing to a broader demographic. A demonstration of our work can be accessed at https://cpp4app.github.io/SeePrivacy/, Comment: USENIX Security 2024. arXiv admin note: text overlap with arXiv:2307.01691

In this paper, we tackle the problem of computing a sequence of rankings with the guarantee of the Pareto-optimal balance between (1) maximizing the utility of the consumers and (2) minimizing unfairness between producers of the items. Such a multi-objective optimization problem is typically solved using a combination of a scalarization method and linear programming on bi-stochastic matrices, representing the distribution of possible rankings of items. However, the above-mentioned approach relies on Birkhoff-von Neumann (BvN) decomposition, of which the computational complexity is $\mathcal{O}(n^5)$ with $n$ being the number of items, making it impractical for large-scale systems. To address this drawback, we introduce a novel approach to the above problem by using the Expohedron - a permutahedron whose points represent all achievable exposures of items. On the Expohedron, we profile the Pareto curve which captures the trade-off between group fairness and user utility by identifying a finite number of Pareto optimal solutions. We further propose an efficient method by relaxing our optimization problem on the Expohedron's circumscribed $n$-sphere, which significantly improve the running time. Moreover, the approximate Pareto curve is asymptotically close to the real Pareto optimal curve as the number of substantial solutions increases. Our methods are applicable with different ranking merits that are non-decreasing functions of item relevance. The effectiveness of our methods are validated through experiments on both synthetic and real-world datasets.

This paper presents a comprehensive overview of the Comparative Opinion Mining from Vietnamese Product Reviews shared task (ComOM), held as part of the 10$^{th}$ International Workshop on Vietnamese Language and Speech Processing (VLSP 2023). The primary objective of this shared task is to advance the field of natural language processing by developing techniques that proficiently extract comparative opinions from Vietnamese product reviews. Participants are challenged to propose models that adeptly extract a comparative "quintuple" from a comparative sentence, encompassing Subject, Object, Aspect, Predicate, and Comparison Type Label. We construct a human-annotated dataset comprising $120$ documents, encompassing $7427$ non-comparative sentences and $2468$ comparisons within $1798$ sentences. Participating models undergo evaluation and ranking based on the Exact match macro-averaged quintuple F1 score., Comment: In Proceedings of VLSP 2023

There has been an increasing interest in physical layer security (PLS), which, compared with conventional cryptography, offers a unique approach to guaranteeing information confidentiality against eavesdroppers. In this paper, we study a joint design of adaptive $M$-ary pulse amplitude modulation (PAM) and precoding, which aims to optimize wiretap visible-light channels' secrecy capacity and bit error rate (BER) performances. The proposed design is motivated by higher-order modulation, which results in better secrecy capacity at the expense of a higher BER. On the other hand, a proper precoding design, which can manipulate the received signal quality at the legitimate user and the eavesdropper, can also enhance secrecy performance and influence the BER. A reward function that considers the secrecy capacity and the BERs of the legitimate user's (Bob) and the eavesdropper's (Eve) channels is introduced and maximized. Due to the non-linearity and complexity of the reward function, it is challenging to solve the optical design using classical optimization techniques. Therefore, reinforcement learning-based designs using Q-learning and Deep Q-learning are proposed to maximize the reward function. Simulation results verify that compared with the baseline designs, the proposed joint designs achieve better reward values while maintaining the BER of Bob's channel (Eve's channel) well below (above) the pre-FEC (forward error correction) BER threshold.

We present an exact static spherisymmetric solution for the Brans-Dicke action sourced by a self-gravitating massless Klein-Gordon helicity-0 field. In contrast to the Maxwell electromagnetic field, a Klein-Gordon field possesses an energy-momentum tensor with $\textit{non-vanishing trace}$. Upon a Weyl mapping into the Einstein frame, the transformed Brans-Dicke scalar field takes on the role of a "dilaton" coupled with the Klein-Gordon field. Despite this dilatonic coupling, the field equations of the resulting Einstein-Klein-Gordon-dilaton action are fully soluble when employing the harmonic radial coordinate. The exact solution derived herein can serve as a prototype for future Brans-Dicke gravity studies involving trace-carrying matter fields. Notably, in the limit of infinite $\omega$, the Brans-Dicke scalar field exhibits an anomalous behavior of ${\cal O}\,(1/\sqrt\omega)$ as opposed to ${\cal O}\,(1/\omega)$. As a consequence, the solution converges to a spacetime configuration of General Relativity sourced by the original Klein-Gordon field and a free scalar field, the latter of which is the ${\cal O}\,(1/\sqrt\omega)$ "remnant" of the Brans-Dicke scalar field. Furthermore, we provide a formal mathematical proof substantiating these two conclusions. Although the ${\cal O}\,(1/\sqrt\omega)$ anomaly has been previously discovered for Brans-Dicke vacuum and Brans-Dicke-Maxwell electrovacuum, our findings establish its prevalence in Brans-Dicke gravity $\textit{regardless}$ of the trace of the energy-momentum tensor of the source. Taken together, the ${\cal O}\,(1/\sqrt\omega)$ anomaly challenges the conventional belief in the ${\cal O}\,(1/\omega)$ signature commonly associated with Brans-Dicke gravity. In particular, it may have implications in improving the relativistic corrections to Newtonian gravity beyond the weak-field parametrized post-Newtonian formalism., Comment: 16 pages, 1 figure. Adding in new Sections VIII and IX

We enable a theory of intrinsic asymptotic expansions for the steady state solutions of the full Navier--Stokes equations. Such a theory was first developed in \cite{FHJ} for Galerkin approximations. To overcome the lack of local compactness in infinite dimensional spaces, we introduce the notion of asymptotic expansions in nested spaces. When the inclusion maps between these spaces are compact, we establish the existence of such an asymptotic expansion for a subsequence of any bounded sequence. This consequently yields an intrinsic asymptotic expansion in a single normed space. We apply this result to the steady states of the Navier--Stokes equations by utilizing the spectral fractional Sobolev spaces. In the case of 2D periodic boundary conditions, more properties relating the terms of the asymptotic expansion are obtained., Comment: 25 pp, submitted for publication

We focus on offline imitation learning (IL), which aims to mimic an expert's behavior using demonstrations without any interaction with the environment. One of the main challenges in offline IL is the limited support of expert demonstrations, which typically cover only a small fraction of the state-action space. While it may not be feasible to obtain numerous expert demonstrations, it is often possible to gather a larger set of sub-optimal demonstrations. For example, in treatment optimization problems, there are varying levels of doctor treatments available for different chronic conditions. These range from treatment specialists and experienced general practitioners to less experienced general practitioners. Similarly, when robots are trained to imitate humans in routine tasks, they might learn from individuals with different levels of expertise and efficiency. In this paper, we propose an offline IL approach that leverages the larger set of sub-optimal demonstrations while effectively mimicking expert trajectories. Existing offline IL methods based on behavior cloning or distribution matching often face issues such as overfitting to the limited set of expert demonstrations or inadvertently imitating sub-optimal trajectories from the larger dataset. Our approach, which is based on inverse soft-Q learning, learns from both expert and sub-optimal demonstrations. It assigns higher importance (through learned weights) to aligning with expert demonstrations and lower importance to aligning with sub-optimal ones. A key contribution of our approach, called SPRINQL, is transforming the offline IL problem into a convex optimization over the space of Q functions. Through comprehensive experimental evaluations, we demonstrate that the SPRINQL algorithm achieves state-of-the-art (SOTA) performance on offline IL benchmarks. Code is available at https://github.com/hmhuy2000/SPRINQL.

Coloring a graph is a well known problem and used in many different contexts. Here we want to assign $k \geq 1$ colors to each vertex of a graph $G$ such that each edge has two different colors at each endpoint. Such a vertex-coloring, if exists, is called a feasible coloring of $G$. \textsc{Distance Coloring} is an extension to the standard \textsc{Coloring} problem. Here we want to enforce that every pair of distinct vertices of distance less than or equal to $d$ have different colors, for integers $d \geq 1$ and $k \geq d+1$. Reconfiguration problems ask if two given configurations can be transformed into each other with certain rules. For example, the well-known \textsc{Coloring Reconfiguration} asks if there is a way to change one vertex's color at a time, starting from a feasible given coloring $\alpha$ of a graph $G$ to reach another feasible given coloring $\beta$ of $G$, such that all intermediate colorings are also feasible. In this paper, we study the reconfiguration of distance colorings on certain graph classes. We show that even for planar, bipartite, and $2$-degenerate graphs, reconfiguring distance colorings is $\mathsf{PSPACE}$-complete for $d \geq 2$ and $k = \Omega(d^2)$ via a reduction from the well-known \textsc{Sliding Tokens} problem. Additionally, we show that the problem on split graphs remains $\mathsf{PSPACE}$-complete when $d = 2$ and large $k$ but can be solved in polynomial time when $d \geq 3$ and $k \geq d+1$, and design a quadratic-time algorithm to solve the problem on paths for any $d \geq 2$ and $k \geq d+1$., Comment: 28 pages, 8 figures, v2: minor revision of v1

Real-world environments are inherently non-stationary, frequently introducing new classes over time. This is especially common in time series classification, such as the emergence of new disease classification in healthcare or the addition of new activities in human activity recognition. In such cases, a learning system is required to assimilate novel classes effectively while avoiding catastrophic forgetting of the old ones, which gives rise to the Class-incremental Learning (CIL) problem. However, despite the encouraging progress in the image and language domains, CIL for time series data remains relatively understudied. Existing studies suffer from inconsistent experimental designs, necessitating a comprehensive evaluation and benchmarking of methods across a wide range of datasets. To this end, we first present an overview of the Time Series Class-incremental Learning (TSCIL) problem, highlight its unique challenges, and cover the advanced methodologies. Further, based on standardized settings, we develop a unified experimental framework that supports the rapid development of new algorithms, easy integration of new datasets, and standardization of the evaluation process. Using this framework, we conduct a comprehensive evaluation of various generic and time-series-specific CIL methods in both standard and privacy-sensitive scenarios. Our extensive experiments not only provide a standard baseline to support future research but also shed light on the impact of various design factors such as normalization layers or memory budget thresholds. Codes are available at https://github.com/zqiao11/TSCIL., Comment: Accepted by SIGKDD 2024 (ADS track). Codes available at https://github.com/zqiao11/TSCIL

This paper proposes a transition system abstraction framework for neural network dynamical system models to enhance the model interpretability, with applications to complex dynamical systems such as human behavior learning and verification. To begin with, the localized working zone will be segmented into multiple localized partitions under the data-driven Maximum Entropy (ME) partitioning method. Then, the transition matrix will be obtained based on the set-valued reachability analysis of neural networks. Finally, applications to human handwriting dynamics learning and verification are given to validate our proposed abstraction framework, which demonstrates the advantages of enhancing the interpretability of the black-box model, i.e., our proposed framework is able to abstract a data-driven neural network model into a transition system, making the neural network model interpretable through verifying specifications described in Computational Tree Logic (CTL) languages., Comment: ACC 2024

In this note we give a computationally easy to use method to compute a maximal extension of certain varieties. As a application we prove that a general paracanonical curve C genus 6 as a codimension three subvarieties of P^4 extend to precisely 26 families of surfaces Y in P^5., Comment: 12 pages

We construct a bijection between certain Deodhar components of a braid variety constructed from an affine Kac-Moody group of type $A_{n-1}$ and vertex-labeled trees on $n$ vertices. By an argument of Galashin, Lam, and Williams using Opdam's trace formula in the affine Hecke algebra and an identity due to Haglund, we obtain an elaborate new proof for the enumeration of the number of vertex-labeled trees on $n$ vertices., Comment: 21 pages

The $k$-Opt algorithm is a local search algorithm for the Traveling Salesman Problem. Starting with an initial tour, it iteratively replaces at most $k$ edges in the tour with the same number of edges to obtain a better tour. Krentel (FOCS 1989) showed that the Traveling Salesman Problem with the $k$-Opt neighborhood is complete for the class PLS (polynomial time local search) and that the $k$-Opt algorithm can have exponential running time for any pivot rule. However, his proof requires $k \gg 1000$ and has a substantial gap. We show the two properties above for a much smaller value of $k$, addressing an open question by Monien, Dumrauf, and Tscheuschner (ICALP 2010). In particular, we prove the PLS-completeness for $k \geq 17$ and the exponential running time for $k \geq 5$., Comment: Appeared in ICALP 2024

The well-known Cluster Vertex Deletion problem (CVD) asks for a given graph $G$ and an integer $k$ whether it is possible to delete a set $S$ of at most $k$ vertices of $G$ such that the resulting graph $G-S$ is a cluster graph (a disjoint union of cliques). We give a complete characterization of graphs $H$ for which CVD on $H$-free graphs is polynomially solvable and for which it is NP-complete. Moreover, in the NP-completeness cases, CVD cannot be solved in sub-exponential time in the vertex number of the $H$-free input graphs unless the Exponential-Time Hypothesis fails. We also consider the connected variant of CVD, the Connected Cluster Vertex Deletion problem (CCVD), in which the set $S$ has to induce a connected subgraph of $G$. It turns out that CCVD admits the same complexity dichotomy for $H$-free graphs. Our results enlarge a list of rare dichotomy theorems for well-studied problems on $H$-free graphs., Comment: Extended version of a MFCS 2022 paper. To appear in Theory of Computing Systems

We present the results of the single-dish observations using the Korean VLBI Network to search for anomalous microwave emission (AME) in nearby galaxies. The targets were selected from MApping the dense moLecular gAs in the sTrongest stAr-formiNg Galaxies (MALATANG), a legacy survey project of the James Clerk Maxwell Telescope. The MALATANG galaxies are good representatives of local galaxies with enhanced nuclear activity associated with star formation and/or AGN, providing IR-bright galaxy samples; thus, they are good candidates for AME hosts. Combining with the ancillary data, we investigated the radio-IR spectral energy distribution (SED), while searching for the AME signals in five galaxies. The AME in NGC 2903 was well detected at a significant confidence level, whereas that in NGC 2146 and M82 was marginal. NGC 1068 and Arp 299 indicated no significant hints, and we provided the upper limits for the AME. The best-fit SED exhibited local peaks of the AME components at higher frequencies and with stronger peak fluxes than those in the previous studies. This suggested the origin of AME being denser environments such as molecular clouds or photodissociation regions rather than warm neutral/ionized medium as commonly suggested by previous studies. Further, our AME-detected targets were observed to exhibit higher specific star-formation rates than the other extragalactic AME hosts. Furthermore, AME favored starburst galaxies among our sample rather than AGN hosts. Consequently, this might imply that AGNs are excessively harsh environments for tiny dust to survive., Comment: 20 pages, 10 figures, accepted for publication in ApJ

In response to the evolving challenges posed by small unmanned aerial vehicles (UAVs), which possess the potential to transport harmful payloads or independently cause damage, we introduce MMAUD: a comprehensive Multi-Modal Anti-UAV Dataset. MMAUD addresses a critical gap in contemporary threat detection methodologies by focusing on drone detection, UAV-type classification, and trajectory estimation. MMAUD stands out by combining diverse sensory inputs, including stereo vision, various Lidars, Radars, and audio arrays. It offers a unique overhead aerial detection vital for addressing real-world scenarios with higher fidelity than datasets captured on specific vantage points using thermal and RGB. Additionally, MMAUD provides accurate Leica-generated ground truth data, enhancing credibility and enabling confident refinement of algorithms and models, which has never been seen in other datasets. Most existing works do not disclose their datasets, making MMAUD an invaluable resource for developing accurate and efficient solutions. Our proposed modalities are cost-effective and highly adaptable, allowing users to experiment and implement new UAV threat detection tools. Our dataset closely simulates real-world scenarios by incorporating ambient heavy machinery sounds. This approach enhances the dataset's applicability, capturing the exact challenges faced during proximate vehicular operations. It is expected that MMAUD can play a pivotal role in advancing UAV threat detection, classification, trajectory estimation capabilities, and beyond. Our dataset, codes, and designs will be available in https://github.com/ntu-aris/MMAUD., Comment: Accepted by ICRA 2024

Work related musculoskeletal disorders (WMSDs) are often caused by repetitive lifting, making them a significant concern in occupational health. Although wearable assist devices have become the norm for mitigating the risk of back pain, most spinal assist devices still possess a partially rigid structure that impacts the user comfort and flexibility. This paper addresses this issue by presenting a smart textile actuated spine assistance robotic exosuit (SARE), which can conform to the back seamlessly without impeding the user movement and is incredibly lightweight. The SARE can assist the human erector spinae to complete any action with virtually infinite degrees of freedom. To detect the strain on the spine and to control the smart textile automatically, a soft knitting sensor which utilizes fluid pressure as sensing element is used. The new device is validated experimentally with human subjects where it reduces peak electromyography (EMG) signals of lumbar erector spinae by around 32 percent in loaded and around 22 percent in unloaded conditions. Moreover, the integrated EMG decreased by around 24.2 percent under loaded condition and around 23.6 percent under unloaded condition. In summary, the artificial muscle wearable device represents an anatomical solution to reduce the risk of muscle strain, metabolic energy cost and back pain associated with repetitive lifting tasks., Comment: 6 pages, 7 figures

Perching with winged Unmanned Aerial Vehicles has often been solved by means of complex control or intricate appendages. Here, we present a simple yet novel method that relies on passive wing morphing for crash-landing on trees and other types of vertical poles. Inspired by the adaptability of animals' and bats' limbs in gripping and holding onto trees, we design dual-purpose wings that enable both aerial gliding and perching on poles. With an upturned nose design, the robot can passively reorient from horizontal flight to vertical upon a head-on crash with a pole, followed by hugging with its wings to perch. We characterize the performance of reorientation and perching in terms of impact speed and angle, pole material, and size. The robot robustly reorients at impact angles above 15{\deg} and speeds of 3 m/s to 9 m/s, and can hold onto various pole types larger than 28% of its wingspan in diameter. We demonstrate crash-perching on tree trunks with an overall success rate of 71%. The method opens up new possibilities for the use of aerial robots in applications such as inspection, maintenance, and biodiversity conservation., Comment: 14 pages, 6 figures. Supplementary material available. Under review at Communications Engineering

Sub-population shift is a specific type of domain shift that highlights changes in data distribution within specific sub-groups or populations between training and testing. Sub-population shift accounts for a significant source of algorithmic bias and calls for distributional robustness. Recent studies found inherent distributional robustness in multi-modality foundation models, such as the vision-language model CLIP, yet this robustness is vulnerable through parameter fine-tuning. In this paper, we propose leveraging the connection of robustness among different modalities and reshaping the distributional robustness of one modality with another. Specifically, in the context of the distributional robustness of CLIP, we propose to leverage natural language inputs to debias the image feature representations, to improve worst-case performance on sub-populations. Our extensive empirical studies show that image representations debiased by natural language can achieve significant performance improvement and reduction of performance instability under sub-population shifts.

This paper introduces the Orthogonal Polynomials Quadrature Algorithm for Survival Analysis (OPSurv), a new method providing time-continuous functional outputs for both single and competing risks scenarios in survival analysis. OPSurv utilizes the initial zero condition of the Cumulative Incidence function and a unique decomposition of probability densities using orthogonal polynomials, allowing it to learn functional approximation coefficients for each risk event and construct Cumulative Incidence Function estimates via Gauss--Legendre quadrature. This approach effectively counters overfitting, particularly in competing risks scenarios, enhancing model expressiveness and control. The paper further details empirical validations and theoretical justifications of OPSurv, highlighting its robust performance as an advancement in survival analysis with competing risks.

High-entropy alloys (HEAs) stand out between multi-component alloys due to their attractive microstructures and mechanical properties. In this investigation, molecular dynamics (MD) simulation and machine learning were used to ascertain the deformation mechanism of AlCoCuCrFeNi HEAs under the influence of temperature, strain rate, and grain sizes. First, the MD simulation shows that the yield stress decreases significantly as the strain and temperature increase. In other cases, changes in strain rate and grain size have less effect on mechanical properties than changes in strain and temperature. The alloys exhibited superplastic behavior under all test conditions. The deformity mechanism discloses that strain and temperature are the main sources of beginning strain, and the shear bands move along the uniaxial tensile axis inside the workpiece. Furthermore, the fast phase shift of inclusion under mild strain indicates the relative instability of the inclusion phase of HCP. Ultimately, the dislocation evolution mechanism shows that the dislocations are transported to free surfaces under increased strain when they nucleate around the grain boundary. Surprisingly, the ML prediction results also confirm the same characteristics as those confirmed from the MD simulation. Hence, the combination of MD and ML reinforces the confidence in the findings of mechanical characteristics of HEA. Consequently, this combination fills the gaps between MD and ML, which can significantly save time human power and cost to conduct real experiments for testing HEA deformation in practice., Comment: It is under revision to submit for a publication

Primal-dual methods have a natural application in Safe Reinforcement Learning (SRL), posed as a constrained policy optimization problem. In practice however, applying primal-dual methods to SRL is challenging, due to the inter-dependency of the learning rate (LR) and Lagrangian multipliers (dual variables) each time an embedded unconstrained RL problem is solved. In this paper, we propose, analyze and evaluate adaptive primal-dual (APD) methods for SRL, where two adaptive LRs are adjusted to the Lagrangian multipliers so as to optimize the policy in each iteration. We theoretically establish the convergence, optimality and feasibility of the APD algorithm. Finally, we conduct numerical evaluation of the practical APD algorithm with four well-known environments in Bullet-Safey-Gym employing two state-of-the-art SRL algorithms: PPO-Lagrangian and DDPG-Lagrangian. All experiments show that the practical APD algorithm outperforms (or achieves comparable performance) and attains more stable training than the constant LR cases. Additionally, we substantiate the robustness of selecting the two adaptive LRs by empirical evidence.

Heat wave (HW) events are becoming more frequent, and they have important consequences because of the negative effects they can have not only on the human population in health terms, but also on biodiversity and agriculture. This motivated a study of the trends in HW events over Extremadura, a region in the southwest of Spain, with much of its area in summer devoted to the production of irrigated crops such as maize and tomatoes. Heat waves were defined for the study as two consecutive days with temperatures above the 95th percentile of the summer (June-August) maximum temperature (Tmax) time series. Two datasets were used: one consisted of 13 daily temperature records uniformly distributed over the Region, and the other was the SPAIN02 gridded observational dataset, extracting just the points corresponding to Extremadura. The trends studied were in the duration, intensity, and frequency of HW events, and in other parameters such as the mean, low (25th percentile), and high (75th percentile) values. In general terms, the results showed significant positive trends in those parameters over the east, the northwest, and a small area in the south of the Region. In order to study changes in HW characteristics (duration, frequency and intensity) considering different subperiods, a stochastic model was used to generate 1000 time series equivalent to the observed ones. The results showed that there were no significant changes in HW duration in the last 10-year subperiod in comparison with the first. But the results were different for warm events (WE), defined with a lower threshold (the 75th percentile), which are also important for agriculture. For several sites, there were significant changes in WE duration, frequency, and intensity., Comment: 36 pages, 11 figures

Pareto Set Learning (PSL) is a promising approach for approximating the entire Pareto front in multi-objective optimization (MOO) problems. However, existing derivative-free PSL methods are often unstable and inefficient, especially for expensive black-box MOO problems where objective function evaluations are costly. In this work, we propose to address the instability and inefficiency of existing PSL methods with a novel controllable PSL method, called Co-PSL. Particularly, Co-PSL consists of two stages: (1) warm-starting Bayesian optimization to obtain quality Gaussian Processes priors and (2) controllable Pareto set learning to accurately acquire a parametric mapping from preferences to the corresponding Pareto solutions. The former is to help stabilize the PSL process and reduce the number of expensive function evaluations. The latter is to support real-time trade-off control between conflicting objectives. Performances across synthesis and real-world MOO problems showcase the effectiveness of our Co-PSL for expensive multi-objective optimization tasks.