Search

This work revisits the classical low-rank matrix factorization problem and unveils the critical role of initialization in shaping convergence rates for such nonconvex and nonsmooth optimization. We introduce Nystrom initialization, which significantly improves the global convergence of Scaled Gradient Descent (ScaledGD) in both symmetric and asymmetric matrix factorization tasks. Specifically, we prove that ScaledGD with Nystrom initialization achieves quadratic convergence in cases where only linear rates were previously known. Furthermore, we extend this initialization to low-rank adapters (LoRA) commonly used for finetuning foundation models. Our approach, NoRA, i.e., LoRA with Nystrom initialization, demonstrates superior performance across various downstream tasks and model scales, from 1B to 7B parameters, in large language and diffusion models.

Sharpness-aware minimization (SAM) improves generalization of various deep learning tasks. Motivated by popular architectures such as LoRA, we explore the implicit regularization of SAM for scale-invariant problems involving two groups of variables. Instead of focusing on commonly used sharpness, this work introduces a concept termed balancedness, defined as the difference between the squared norm of two variables. This allows us to depict richer global behaviors of SAM. In particular, our theoretical and empirical findings reveal that i) SAM promotes balancedness; and ii) the regularization on balancedness is data-responsive -- outliers have stronger impact. The latter coincides with empirical observations that SAM outperforms SGD in the presence of outliers. Leveraging the implicit regularization, we develop a resource-efficient SAM variant, balancedness-aware regularization (BAR), tailored for scale-invariant problems such as finetuning language models with LoRA. BAR saves 95% computational overhead of SAM, with enhanced test performance across various tasks on RoBERTa, GPT2, and OPT-1.3B., Comment: NeurIPS 2024

Quantum ant colony optimization (QACO) has drew much attention since it combines the advantages of quantum computing and ant colony optimization (ACO) algorithm overcoming some limitations of the traditional ACO algorithm. However,due to the hardware resource limitations of currently available quantum computers, the practical application of the QACO is still not realized. In this paper, we developed a quantum-classical hybrid algorithm by combining the clustering algorithm with QACO algorithm.This extended QACO can handle large-scale optimization problems with currently available quantum computing resource. We have tested the effectiveness and performance of the extended QACO algorithm with the Travelling Salesman Problem (TSP) as benchmarks, and found the algorithm achieves better performance under multiple diverse datasets. In addition, we investigated the noise impact on the extended QACO and evaluated its operation possibility on current available noisy intermediate scale quantum(NISQ) devices. Our work shows that the combination of the clustering algorithm with QACO effectively improved its problem solving scale, which makes its practical application possible in current NISQ era of quantum computing., Comment: arXiv admin note: substantial text overlap with arXiv:2403.00367

Keyphrase generation (KPG) aims to automatically generate a collection of phrases representing the core concepts of a given document. The dominant paradigms in KPG include one2seq and one2set. Recently, there has been increasing interest in applying large language models (LLMs) to KPG. Our preliminary experiments reveal that it is challenging for a single model to excel in both recall and precision. Further analysis shows that: 1) the one2set paradigm owns the advantage of high recall, but suffers from improper assignments of supervision signals during training; 2) LLMs are powerful in keyphrase selection, but existing selection methods often make redundant selections. Given these observations, we introduce a generate-then-select framework decomposing KPG into two steps, where we adopt a one2set-based model as generator to produce candidates and then use an LLM as selector to select keyphrases from these candidates. Particularly, we make two important improvements on our generator and selector: 1) we design an Optimal Transport-based assignment strategy to address the above improper assignments; 2) we model the keyphrase selection as a sequence labeling task to alleviate redundant selections. Experimental results on multiple benchmark datasets show that our framework significantly surpasses state-of-the-art models, especially in absent keyphrase prediction., Comment: Accepted by EMNLP 2024 Main Conference

In this paper, we investigate the semantic clustering properties of deep reinforcement learning (DRL) for video games, enriching our understanding of the internal dynamics of DRL and advancing its interpretability. In this context, semantic clustering refers to the inherent capacity of neural networks to internally group video inputs based on semantic similarity. To achieve this, we propose a novel DRL architecture that integrates a semantic clustering module featuring both feature dimensionality reduction and online clustering. This module seamlessly integrates into the DRL training pipeline, addressing instability issues observed in previous t-SNE-based analysis methods and eliminating the necessity for extensive manual annotation of semantic analysis. Through experiments, we validate the effectiveness of the proposed module and the semantic clustering properties in DRL for video games. Additionally, based on these properties, we introduce new analytical methods to help understand the hierarchical structure of policies and the semantic distribution within the feature space.

We use the rms and lag spectra of the type-C quasi-periodic oscillation (QPO) to study the properties of the Comptonisation region (aka corona) during the low/hard and hard-intermediate states of the main outburst and reflare of MAXI J1348$-$630. We simultaneously fit the time-averaged energy spectrum of the source and the fractional rms and phase-lag spectra of the QPO with the time-dependent Comptonization model vKompth. The data can be explained by two physically connected coronae interacting with the accretion disc via a feedback loop of X-ray photons. The best-fitting model consists of a corona of $\sim$10$^3$ km located at the inner edge of the disc and a second corona of $\sim$10$^4$ km horizontally extended and covering the inner parts of the accretion disc. The properties of both coronae during the reflare are similar to those during the low/hard state of the main outburst, reinforcing the idea that both the outburst and the reflare are driven by the same physical mechanisms. We combine our results for the type-C QPO with those from previous work focused on the study of type-A and type-B QPOs with the same model to study the evolution of the geometry of the corona through the whole outburst, including the reflare of MAXI J1348$-$630. Finally, we show that the sudden increase in the phase-lag frequency spectrum and the sharp drop in the coherence function previously observed in MAXI J1348$-$630 are due to the type-C QPO during the decay of the outburst and can be explained in terms of the geometry of the coronae., Comment: 18 pages, 8 figures, 1 table. Submitted to ApJ

Learning performance data describe correct and incorrect answers or problem-solving attempts in adaptive learning, such as in intelligent tutoring systems (ITSs). Learning performance data tend to be highly sparse (80\%\(\sim\)90\% missing observations) in most real-world applications due to adaptive item selection. This data sparsity presents challenges to using learner models to effectively predict future performance explore new hypotheses about learning. This article proposes a systematic framework for augmenting learner data to address data sparsity in learning performance data. First, learning performance is represented as a three-dimensional tensor of learners' questions, answers, and attempts, capturing longitudinal knowledge states during learning. Second, a tensor factorization method is used to impute missing values in sparse tensors of collected learner data, thereby grounding the imputation on knowledge tracing tasks that predict missing performance values based on real observations. Third, a module for generating patterns of learning is used. This study contrasts two forms of generative Artificial Intelligence (AI), including Generative Adversarial Networks (GANs) and Generate Pre-Trained Transformers (GPT) to generate data associated with different clusters of learner data. We tested this approach on an adult literacy dataset from AutoTutor lessons developed for Adult Reading Comprehension (ARC). We found that: (1) tensor factorization improved the performance in tracing and predicting knowledge mastery compared with other knowledge tracing techniques without data augmentation, showing higher relative fidelity for this imputation method, and (2) the GAN-based simulation showed greater overall stability and less statistical bias based on a divergence evaluation with varying simulation sample sizes compared to GPT.

Multimodel Large Language Models(MLLMs) have achieved promising OCR-free Document Understanding performance by increasing the supported resolution of document images. However, this comes at the cost of generating thousands of visual tokens for a single document image, leading to excessive GPU memory and slower inference times, particularly in multi-page document comprehension. In this work, to address these challenges, we propose a High-resolution DocCompressor module to compress each high-resolution document image into 324 tokens, guided by low-resolution global visual features. With this compression module, to strengthen multi-page document comprehension ability and balance both token efficiency and question-answering performance, we develop the DocOwl2 under a three-stage training framework: Single-image Pretraining, Multi-image Continue-pretraining, and Multi-task Finetuning. DocOwl2 sets a new state-of-the-art across multi-page document understanding benchmarks and reduces first token latency by more than 50%, demonstrating advanced capabilities in multi-page questioning answering, explanation with evidence pages, and cross-page structure understanding. Additionally, compared to single-image MLLMs trained on similar data, our DocOwl2 achieves comparable single-page understanding performance with less than 20% of the visual tokens. Our codes, models, and data are publicly available at https://github.com/X-PLUG/mPLUG-DocOwl/tree/main/DocOwl2., Comment: 15 pages, 7 figures

While the classic Prospect Theory has highlighted the reference-dependent and comparative nature of consumers' product evaluation processes, few models have successfully integrated this theoretical hypothesis into data-driven preference quantification, particularly in the realm of recommender systems development. To bridge this gap, we propose a new research problem of modeling reference-dependent preferences from a data-driven perspective, and design a novel deep learning-based framework named Attributed Reference-dependent Choice Model for Recommendation (ArcRec) to tackle the inherent challenges associated with this problem. ArcRec features in building a reference network from aggregated historical purchase records for instantiating theoretical reference points, which is then decomposed into product attribute specific sub-networks and represented through Graph Neural Networks. In this way, the reference points of a consumer can be encoded at the attribute-level individually from her past experiences but also reflect the crowd influences. ArcRec also makes novel contributions to quantifying consumers' reference-dependent preferences using a deep neural network-based utility function that integrates both interest-inspired and price-inspired preferences, with their complex interaction effects captured by an attribute-aware price sensitivity mechanism. Most importantly, ArcRec introduces a novel Attribute-level Willingness-To-Pay measure to the reference-dependent utility function, which captures a consumer's heterogeneous salience of product attributes via observing her attribute-level price tolerance to a product. Empirical evaluations on both synthetic and real-world online shopping datasets demonstrate ArcRec's superior performances over fourteen state-of-the-art baselines.

A quantum network provides an infrastructure connecting quantum devices with revolutionary computing, sensing, and communication capabilities. As the best-known application of a quantum network, quantum key distribution (QKD) shares secure keys guaranteed by the laws of quantum mechanics. A quantum satellite constellation offers a solution to facilitate the quantum network on a global scale. The Micius satellite has verified the feasibility of satellite quantum communications, however, scaling up quantum satellite constellations is challenging, requiring small lightweight satellites, portable ground stations and real-time secure key exchange. Here we tackle these challenges and report the development of a quantum microsatellite capable of performing space-to-ground QKD using portable ground stations. The quantum microsatellite features a payload weighing approximately 23 kg, while the portable ground station weighs about 100 kg. These weights represent reductions by more than an order and two orders of magnitude, respectively, compared to the Micius satellite. Additionally, we multiplex bidirectional satellite-ground optical communication with quantum communication, enabling key distillation and secure communication in real-time. Using the microsatellite and the portable ground stations, we demonstrate satellite-based QKD with multiple ground stations and achieve the sharing of up to 0.59 million bits of secure keys during a single satellite pass. The compact quantum payload can be readily assembled on existing space stations or small satellites, paving the way for a satellite-constellation-based quantum and classical network for widespread real-life applications., Comment: 40 pages, 8 figures

We present a comprehensive study based on multi-wavelength observations from the NuSTAR, NICER, Swift, Fermi, NEOWISE, and ATCA telescopes during the 2022 outburst of the black hole X-ray binary IGR J17091-3624. Our investigation concentrates on the heartbeat-like variability in the X-ray emission, with the aim of using it as a tool to unravel the origin of the non-thermal emission during the heartbeat state. Through X-ray timing and spectral analysis, we observe that the heartbeat-like variability correlates with changes in the disk temperature, supporting the disk radiation pressure instability scenario. Moreover, in addition to a Comptonization component, our time-averaged and phase-resolved spectroscopy reveal the presence of a power-law component that varies independently from the disk component. Combined with the radio to X-ray spectral energy distribution fitting, our results suggest that the power-law component could originate from synchrotron self-Compton radiation in the jet, which requires a strong magnetic field of about $B = (0.3$-$3.5)\times10^6$ G. Additionally, assuming that IGR J17091-3624 and GRS 1915+105 share the same radio-X-ray correlation coefficient during both the hard and the heartbeat states, we obtain a distance of $13.7\pm2.3$ kpc for IGR J17091-3624., Comment: 19 pages, 11 figures, 3 tables; accepted for publication in ApJ

Quantum computing combined with machine learning (ML) is an extremely promising research area, with numerous studies demonstrating that quantum machine learning (QML) is expected to solve scientific problems more effectively than classical ML. In this work, we successfully apply QML to drug discovery, showing that QML can significantly improve model performance and achieve faster convergence compared to classical ML. Moreover, we demonstrate that the model accuracy of the QML improves as the number of qubits increases. We also introduce noise to the QML model and find that it has little effect on our experimental conclusions, illustrating the high robustness of the QML model. This work highlights the potential application of quantum computing to yield significant benefits for scientific advancement as the qubit quantity increase and quality improvement in the future.

Learning performance data, such as correct or incorrect responses to questions in Intelligent Tutoring Systems (ITSs) is crucial for tracking and assessing the learners' progress and mastery of knowledge. However, the issue of data sparsity, characterized by unexplored questions and missing attempts, hampers accurate assessment and the provision of tailored, personalized instruction within ITSs. This paper proposes using the Generative Adversarial Imputation Networks (GAIN) framework to impute sparse learning performance data, reconstructed into a three-dimensional (3D) tensor representation across the dimensions of learners, questions and attempts. Our customized GAIN-based method computational process imputes sparse data in a 3D tensor space, significantly enhanced by convolutional neural networks for its input and output layers. This adaptation also includes the use of a least squares loss function for optimization and aligns the shapes of the input and output with the dimensions of the questions-attempts matrices along the learners' dimension. Through extensive experiments on six datasets from various ITSs, including AutoTutor, ASSISTments and MATHia, we demonstrate that the GAIN approach generally outperforms existing methods such as tensor factorization and other generative adversarial network (GAN) based approaches in terms of imputation accuracy. This finding enhances comprehensive learning data modeling and analytics in AI-based education.

Low-frequency quasi-periodic oscillations (LFQPOs) are commonly observed in X-ray light curves of black hole X-ray binaries (BHXRBs); however, their origin remains a topic of debate. In order to thoroughly investigate variations in spectral properties on the QPO timescale, we utilized the Hilbert-Huang transform technique to conduct phase-resolved spectroscopy across a broad energy band for LFQPOs in the newly discovered BHXRB Swift J1727.8-1613. This is achieved through quasi-simultaneous observations from Neutron star Interior Composition ExploreR (NICER), Nuclear Spectroscopic Telescope ARray (NuSTAR), and Hard X-ray Modulation Telescope (Insight-HXMT). Our analysis reveals that both the non-thermal and disk-blackbody components exhibit variations on the QPO timescale, with the former dominating the QPO variability. For the spectral parameters, we observe modulation of the disk temperature, spectral indices, and reflection fraction with the QPO phase with high statistical significance (>5\sigma). Notably, the variation in the disk temperature is found to precede the variations in the non-thermal and disk fluxes by ~0.4-0.5 QPO cycles. We suggest that these findings offer further evidence that the type-C QPO variability is a result of geometric effects of the accretion flow., Comment: Accepted for pulication in The Astrophysical Journal

Despite the success of Heterogeneous Graph Neural Networks (HGNNs) in modeling real-world Heterogeneous Information Networks (HINs), challenges such as expressiveness limitations and over-smoothing have prompted researchers to explore Graph Transformers (GTs) for enhanced HIN representation learning. However, research on GT in HINs remains limited, with two key shortcomings in existing work: (1) A node's neighbors at different distances in HINs convey diverse semantics. Unfortunately, existing methods ignore such differences and uniformly treat neighbors within a given distance in a coarse manner, which results in semantic confusion. (2) Nodes in HINs have various types, each with unique semantics. Nevertheless, existing methods mix nodes of different types during neighbor aggregation, hindering the capture of proper correlations between nodes of diverse types. To bridge these gaps, we design an innovative structure named (k,t)-ring neighborhood, where nodes are initially organized by their distance, forming different non-overlapping k-ring neighborhoods for each distance. Within each k-ring structure, nodes are further categorized into different groups according to their types, thus emphasizing the heterogeneity of both distances and types in HINs naturally. Based on this structure, we propose a novel Hierarchical Heterogeneous Graph Transformer (HHGT) model, which seamlessly integrates a Type-level Transformer for aggregating nodes of different types within each k-ring neighborhood, followed by a Ring-level Transformer for aggregating different k-ring neighborhoods in a hierarchical manner. Extensive experiments are conducted on downstream tasks to verify HHGT's superiority over 14 baselines, with a notable improvement of up to 24.75% in NMI and 29.25% in ARI for node clustering task on the ACM dataset compared to the best baseline.

Relativistic jets are thought to play a crucial role in the formation of massive galaxies and supermassive black holes. Here we report multi-wavelength and multi-epoch observations of the quasar VLASSJ0410-0139 at redshift z=7, powered by a 7e8 solar-mass black hole. Its radio variability, X-ray properties, and compact radio emission on parsec scales reveal that J0410-0139 is a blazar with a relativistic jet aligned with our line of sight. This blazar's existence implies that many more similar (unaligned) jetted sources must exist at z=7. One scenario is that we observe an intrinsically low-power radio jet, but we see it at high luminosity due to relativistic beaming effects. In this case, a large fraction (>80%) of the UV bright quasars must have a similar jet to match the number density expected from the UV quasar luminosity function. These jets can enhance the growth of supermassive black holes and substantially affect their host galaxies. However, the implications would be even more severe if the quasar belongs to the top 10% radio luminous quasars, as measured if the beaming enhancement is less than a factor of 10-15. In this scenario, there should be hundreds to thousands of radio-quiet quasars at z=7 with intrinsic properties similar to J0410-0139 -- in strong tension with the number density of bright quasars derived from their UV luminosity function. To reconcile these results, most black hole growth at z=7 must happen in an obscured phase, as some models predict. The existence of supermassive black holes in the epoch of reionization is facilitated by significant jet-enhanced or obscured super-Eddington accretion., Comment: Submitted

We present an energy-dependent analysis for the type-C quasi-periodic oscillations (QPOs) observed in the black hole X-ray binary Swift J1727.8-1613 using Insight-HXMT observations. We find that the QPO fractional rms at energies above 40 keV is significantly higher than that below 20 keV. This is the first report of a high energy (HE)-rms excess in the rms spectrum of a black hole X-ray binary. In the high energy band, an extra hard component is observed in additional to the standard thermal Comptonization component at similar energy band. The value of the QPO HE-rms excess is not only correlated with the disk parameters and the photon index of the standard Comptonization component, but also exhibits a moderate positive correlation with the flux of the additional hard spectral component. No features in the QPO phase-lag spectra are seen corresponding to the additional hard component. We propose that the additional hard component in the spectrum may originate from jet emission and the associated QPO HE-rms excess can be explained by the precession of the jet base.

Neuromorphic sensors, specifically event cameras, revolutionize visual data acquisition by capturing pixel intensity changes with exceptional dynamic range, minimal latency, and energy efficiency, setting them apart from conventional frame-based cameras. The distinctive capabilities of event cameras have ignited significant interest in the domain of event-based action recognition, recognizing their vast potential for advancement. However, the development in this field is currently slowed by the lack of comprehensive, large-scale datasets, which are critical for developing robust recognition frameworks. To bridge this gap, we introduces DailyDVS-200, a meticulously curated benchmark dataset tailored for the event-based action recognition community. DailyDVS-200 is extensive, covering 200 action categories across real-world scenarios, recorded by 47 participants, and comprises more than 22,000 event sequences. This dataset is designed to reflect a broad spectrum of action types, scene complexities, and data acquisition diversity. Each sequence in the dataset is annotated with 14 attributes, ensuring a detailed characterization of the recorded actions. Moreover, DailyDVS-200 is structured to facilitate a wide range of research paths, offering a solid foundation for both validating existing approaches and inspiring novel methodologies. By setting a new benchmark in the field, we challenge the current limitations of neuromorphic data processing and invite a surge of new approaches in event-based action recognition techniques, which paves the way for future explorations in neuromorphic computing and beyond. The dataset and source code are available at https://github.com/QiWang233/DailyDVS-200., Comment: Accepted to ECCV 2024

Building Management System (BMS) through a data-driven method always faces data and model scalability issues. We propose a methodology to tackle the scalability challenges associated with the development of data-driven models for BMS by using Large Language Models (LLMs). LLMs' code generation adaptability can enable broader adoption of BMS by "automating the automation," particularly the data handling and data-driven modeling processes. In this paper, we use LLMs to generate code that processes structured data from BMS and build data-driven models for BMS's specific requirements. This eliminates the need for manual data and model development, reducing the time, effort, and cost associated with this process. Our hypothesis is that LLMs can incorporate domain knowledge about data science and BMS into data processing and modeling, ensuring that the data-driven modeling is automated for specific requirements of different building types and control objectives, which also improves accuracy and scalability. We generate a prompt template following the framework of Machine Learning Operations so that the prompts are designed to systematically generate Python code for data-driven modeling. Our case study indicates that bi-sequential prompting under the prompt template can achieve a high success rate of code generation and code accuracy, and significantly reduce human labor costs.

Cross-document Relation Extraction aims to predict the relation between target entities located in different documents. In this regard, the dominant models commonly retain useful information for relation prediction via bridge entities, which allows the model to elaborately capture the intrinsic interdependence between target entities. However, these studies ignore the non-bridge entities, each of which co-occurs with only one target entity and offers the semantic association between target entities for relation prediction. Besides, the commonly-used dataset--CodRED contains substantial NA instances, leading to the prediction bias during inference. To address these issues, in this paper, we propose a novel graph-based cross-document RE model with non-bridge entity enhancement and prediction debiasing. Specifically, we use a unified entity graph to integrate numerous non-bridge entities with target entities and bridge entities, modeling various associations between them, and then use a graph recurrent network to encode this graph. Finally, we introduce a novel debiasing strategy to calibrate the original prediction distribution. Experimental results on the closed and open settings show that our model significantly outperforms all baselines, including the GPT-3.5-turbo and InstructUIE, achieving state-of-the-art performance. Particularly, our model obtains 66.23% and 55.87% AUC points in the official leaderboard\footnote{\url{https://codalab.lisn.upsaclay.fr/competitions/3770#results}} under the two settings, respectively, ranking the first place in all submissions since December 2023. Our code is available at https://github.com/DeepLearnXMU/CoRE-NEPD., Comment: Accepted to ACL 2024 Findings

In category learning, a growing body of literature has increasingly focused on exploring the impacts of interleaving in contrast to blocking. The sequential attention hypothesis posits that interleaving draws attention to the differences between categories while blocking directs attention toward similarities within categories. Although a recent study underscores the joint influence of memory and attentional factors on sequencing effects, there remains a scarcity of effective computational models integrating both attentional and memory considerations to comprehensively understand the effect of training sequences on students' performance. This study introduces a novel integration of attentional factors and spacing into the logistic knowledge tracing (LKT) models to monitor students' performance across different training sequences (interleaving and blocking). Attentional factors were incorporated by recording the counts of comparisons between adjacent trials, considering whether they belong to the same or different category. Several features were employed to account for temporal spacing. We used cross-validations to test the model fit and predictions on the learning session and posttest. Our findings reveal that incorporating both attentional factors and spacing features in the Additive Factors Model (AFM) significantly enhances its capacity to capture the effects of interleaving and blocking and demonstrates superior predictive accuracy for students' learning outcomes. By bridging the gap between attentional factors and memory processes, our computational approach offers a more comprehensive framework for understanding and predicting category learning outcomes in educational settings., Comment: 7 pages, 3 figures, Educational Data Mining 2024

Dialogue-based Intelligent Tutoring Systems (ITSs) have significantly advanced adaptive and personalized learning by automating sophisticated human tutoring strategies within interactive dialogues. However, replicating the nuanced patterns of expert human communication remains a challenge in Natural Language Processing (NLP). Recent advancements in NLP, particularly Large Language Models (LLMs) such as OpenAI's GPT-4, offer promising solutions by providing human-like and context-aware responses based on extensive pre-trained knowledge. Motivated by the effectiveness of LLMs in various educational tasks (e.g., content creation and summarization, problem-solving, and automated feedback provision), our study introduces the Socratic Playground for Learning (SPL), a dialogue-based ITS powered by the GPT-4 model, which employs the Socratic teaching method to foster critical thinking among learners. Through extensive prompt engineering, SPL can generate specific learning scenarios and facilitates efficient multi-turn tutoring dialogues. The SPL system aims to enhance personalized and adaptive learning experiences tailored to individual needs, specifically focusing on improving critical thinking skills. Our pilot experimental results from essay writing tasks demonstrate SPL has the potential to improve tutoring interactions and further enhance dialogue-based ITS functionalities. Our study, exemplified by SPL, demonstrates how LLMs enhance dialogue-based ITSs and expand the accessibility and efficacy of educational technologies.

We report data analysis results about the outburst evolution and spectral properties during the hard state of the recently discovered X-ray transient Swift J1727.8-163 as observed by \emph{Insight}-HXMT and NuSTAR. We find that the broadband X-ray spectrum of Swift J1727.8-163 is more complex than the most typical spectral patterns of black hole X-ray binary systems, with not only a comparatively weaker reflection component but also an additional spectral continuum component, manifesting itself as a hard X-ray tail beyond the thermal Comptonization description detectable below 100 keV. This additional component can be phenomenologically well fitted by adding an extra power-law model with high energy exponential cutoff in the 2-120 keV energy band. We made an attempt to explain the broadband X-ray spectral continuum with a thermal/non-thermal hybrid plasma corona scenario , and find an ultra high compactness parameter ($l_{\rm s}\sim2000$) and a steep non-thermal electron distribution ($\Gamma_{\rm inj}>4$), suggesting the source was accreting with high Eddington rates and that the electron acceleration mechanism is not very efficient. We also present a detailed multi-epoch analysis of spectral properties using \emph{Insight}-HXMT data to investigate the evolution of the key physical properties regarding the disk and corona during the hard states. No significant variation is found with the inner disk radius and the coronal temperature during this time period, and the weak reflection and hard X-ray tail features are persistent. We discuss the physical implications of our spectral analysis results in the context of disk-corona relation, particle acceleration, and jet contribution, during the rise of a black hole X-ray binary in outburst., Comment: 16 pages, 6 figures

We present the discovery of X-ray polarization from the neutron star low-mass X-ray binary and Z-source, GX~340$+$0, using an Imaging X-ray Polarimetry Explorer (IXPE) observation in March 2024. Along with the IXPE observation, we conducted an extensive X-ray and radio monitoring campaign to ascertain the source properties during and around the IXPE observation. The source was within the horizontal branch throughout the multiwavelength campaign. We measured a significant X-ray polarization in 2--8 keV with polarization degree (PD) = $4.02 \pm 0.35$% and polarization angle (PA) = $37.6 \pm 2.5^\circ$. The energy-dependent polarization indicates that in the 2-2.5 keV energy range, the PA is much lower, $\sim9\pm8^\circ$, while other energy bands are consistent with the PA found over 2.5--8 keV. The simultaneous AstroSat-IXPE spectro-polarimetric observations provide some evidence for independent polarization from various spectral components, hinting at a disparity in the PA from the accretion disk and the Comptonized emission, while suggesting an unpolarized emission from the blackbody component. Radio observations in the 0.7--9 GHz frequency range reveal a non-detection of radio emission in 0.7-1.5 GHz and a significant detection in 5.5--9 GHz, suggesting the presence of a spectral break in 1.5-5.5 GHz. Using ATCA observation we place upper limits on the radio polarization at $<$6% on the linear polarization and $<$4% on the circular polarization at 3$\sigma$ level. We discuss the origin of the X-ray polarization and its implications on the geometry of the spectral components., Comment: Submitted in ApJL, 4 figures, 3 tables

Continuous-time approximation of Stochastic Gradient Descent (SGD) is a crucial tool to study its escaping behaviors from stationary points. However, existing stochastic differential equation (SDE) models fail to fully capture these behaviors, even for simple quadratic objectives. Built on a novel stochastic backward error analysis framework, we derive the Hessian-Aware Stochastic Modified Equation (HA-SME), an SDE that incorporates Hessian information of the objective function into both its drift and diffusion terms. Our analysis shows that HA-SME matches the order-best approximation error guarantee among existing SDE models in the literature, while achieving a significantly reduced dependence on the smoothness parameter of the objective. Further, for quadratic objectives, under mild conditions, HA-SME is proved to be the first SDE model that recovers exactly the SGD dynamics in the distributional sense. Consequently, when the local landscape near a stationary point can be approximated by quadratics, HA-SME is expected to accurately predict the local escaping behaviors of SGD.

Mathematical modeling of production systems is the foundation of all model-based approaches for production system analysis, design, improvement, and control. To construct such a model for the stochastic process of the production system more efficiently, a new modeling approach has been proposed that reversely identifies the model parameters using system performance metrics (e.g., production rate, work-in-process, etc.) derived from the parts flow data. This paper extends this performance metrics-based modeling approach to non-exponential serial production lines. Since no analytical expressions of performance metrics are available for non-exponential systems, we use neural network surrogate models to calculate those performance metrics as functions in terms of the system parameters. Then, based on the surrogate models and given performance metrics, the machine parameters are estimated by solving a constrained optimization problem that minimizes the mean square error of the performance metrics resulting from the estimated parameters compared to the true ones. With the designed multi-start particle swarm optimization algorithm, we find that multiple non-unique combinations of machine parameters can lead to practically the same system performance metrics and a linear relationship of the reliability parameters from these obtained estimations is observed. Besides, model sensitivity analysis is implemented to verify the robustness of the different combinations of machine parameters even under the potential improvement scenarios., Comment: 31 pages

Emerging Artificial Intelligence of Things (AIoT) applications desire online prediction using deep neural network (DNN) models on mobile devices. However, due to the movement of devices, unfamiliar test samples constantly appear, significantly affecting the prediction accuracy of a pre-trained DNN. In addition, unstable network connection calls for local model inference. In this paper, we propose a light-weight scheme, called Anole, to cope with the local DNN model inference on mobile devices. The core idea of Anole is to first establish an army of compact DNN models, and then adaptively select the model fitting the current test sample best for online inference. The key is to automatically identify model-friendly scenes for training scene-specific DNN models. To this end, we design a weakly-supervised scene representation learning algorithm by combining both human heuristics and feature similarity in separating scenes. Moreover, we further train a model classifier to predict the best-fit scene-specific DNN model for each test sample. We implement Anole on different types of mobile devices and conduct extensive trace-driven and real-world experiments based on unmanned aerial vehicles (UAVs). The results demonstrate that Anole outwits the method of using a versatile large DNN in terms of prediction accuracy (4.5% higher), response time (33.1% faster) and power consumption (45.1% lower).

We conduct a comprehensive analysis of the accreting X-ray pulsar, Hercules X-1, utilizing data from IXPE and NuSTAR. IXPE performed five observations of Her X-1, consisting of three in the Main-on state and two in the Short-on state. Our time-resolved analysis uncovers the linear correlations between the flux and polarization degree as well as the pulse fraction and polarization degree. Geometry parameters are rigorously constrained by fitting the phase-resolved modulations of Cyclotron Resonance Scattering Feature and polarization angle with a simple dipole model and Rotating Vector Model respectively, yielding roughly consistent results. The changes of $\chi_{\rm p}$ (the position angle of the pulsar's spin axis on the plane of the sky) between different Main-on observations suggest the possible forced precession of the neutron star crust. Furthermore, a linear association between the energy of Cyclotron Resonance Scattering Feature and polarization angle implies the prevalence of a dominant dipole magnetic field, and their phase-resolved modulations likely arise from viewing angle effects., Comment: Accepted for MNRAS

Charts are important for presenting and explaining complex data relationships. Recently, multimodal large language models (MLLMs) have shown remarkable capabilities in various chart understanding tasks. However, the sheer size of these models in terms of parameters and computational requirements limits their use in resource-constrained environments. In this paper, we present TinyChart, an efficient MLLM for chart understanding with only 3B parameters. TinyChart overcomes two key challenges in efficient chart understanding: (1) reduce the burden of learning numerical computations through a Program-of-Thoughts (PoT) learning strategy, which trains the model to generate Python programs for numerical calculations, and (2) reduce lengthy vision feature sequences produced by the vision transformer for high-resolution images through a Vision Token Merging module, which gradually merges most similar vision tokens. Extensive experiments demonstrate that our 3B TinyChart achieves SOTA performance on a variety of chart understanding benchmarks including ChartQA, Chart-to-Text, Chart-to-Table, OpenCQA, and ChartX. It outperforms several chart understanding MLLM with up to 13B parameters such as ChartLlama and ChartAst, and close-sourced general-purpose MLLM GPT-4V on ChartQA. It also demonstrates its superior efficiency with higher throughput during inference due to a smaller model scale and more efficient vision encoding. Our code and model are available at https://github.com/X-PLUG/mPLUG-DocOwl/tree/main/TinyChart., Comment: 13 pages, 11 figures

Millisecond pulsars (MSPs) are known for their long-term stability. Using six years of observations from the Neutron Star Interior Composition Explorer (NICER), we have conducted an in-depth analysis of the X-ray timing results for six MSPs: PSRs B1937+21, B1821$-$24, J0437$-$4715, J0030+0451, J0218+4232, and J2124$-$3358. The timing stability parameter $\sigma_z$ has been calculated, revealing remarkable timing precision on the order of $10^{-14}$ for PSRs B1937+21 and J0437$-$4715, and $10^{-13}$ for PSRs B1821$-$24, J0218+4232, and J0030+0451 over a timescale of 1000 days. These findings underscore the feasibility of autonomous in-orbit timekeeping using X-ray observations of MSPs. In addition, the consistency of long-term spin-down noise in the X-ray and radio bands has been investigated by comparison with IPTA radio data.

Two-color laser induced plasma filamentation in air could serve as tabletop sources of broadband terahertz (THz) pulses. Ubiquitous air in the earth facilitates its widespread utilities, particularly, in wireless communication and remote sensing, exploiting the unique advantage of the air-based scheme that the THz sources can be delivered over standoff distances via the pump laser propagation in air. However, the THz emission pattern inevitably has a conical angular profile with a dip in the propagation axis, therefore, THz energy concentration, propagation directionality, and accuracy of signal demodulation are significantly impaired, greatly limiting its direct applications. Here, we successfully eliminate the unfavorable conical profile by experiments and meanwhile enhance THz directionality and intensity by 17 folds by use of abruptly-autofocusing laser beams. Our theory and simulations show that these observations are attributed to efficient suppression of the dephasing effect appearing in previous investigations with ordinary Gaussian laser beams. This scheme is easily accessible since the abruptly-autofocusing beam can be achieved by imposing a spatial light modulator on the input Gaussian beam. This study solves a long-standing problem in the two-color laser scheme and clarifies the underlying physics of the conical angular profile formation., Comment: 22 pages, 4 figures, 1 table

The generation and collection of big data series are becoming an integral part of many emerging applications in sciences, IoT, finance, and web applications among several others. The terabyte-scale of data series has motivated recent efforts to design fully distributed techniques for supporting operations such as approximate kNN similarity search, which is a building block operation in most analytics services on data series. Unfortunately, these techniques are heavily geared towards achieving scalability at the cost of sacrificing the results' accuracy. State-of-the-art systems report accuracy below 10% and 40%, respectively, which is not practical for many real-world applications. In this paper, we investigate the root problems in these existing techniques that limit their ability to achieve better a trade-off between scalability and accuracy. Then, we propose a framework, called CLIMBER, that encompasses a novel feature extraction mechanism, indexing scheme, and query processing algorithms for supporting approximate similarity search in big data series. For CLIMBER, we propose a new loss-resistant dual representation composed of rank-sensitive and ranking-insensitive signatures capturing data series objects. Based on this representation, we devise a distributed two-level index structure supported by an efficient data partitioning scheme. Our similarity metrics tailored for this dual representation enables meaningful comparison and distance evaluation between the rank-sensitive and ranking-insensitive signatures. Finally, we propose two efficient query processing algorithms, CLIMBER-kNN and CLIMBER-kNN-Adaptive, for answering approximate kNN similarity queries. Our experimental study on real-world and benchmark datasets demonstrates that CLIMBER, unlike existing techniques, features results' accuracy above 80% while retaining the desired scalability to terabytes of data., Comment: 16 pages, 14 figures, 1 table

Recently, the concept of generating function has been employed in one-loop reduction. For one-loop integrals encompassing arbitrary tensor ranks and higher-pole contributions, the generating function can be decomposed into a tensor part and a higher-pole part. While the tensor component has been thoroughly addressed in recent studies, there remains a lack of satisfactory investigations regarding the higher-pole part. In this work, we completely solve the problem. We first establish the partial differential equations governing the higher-pole generating function. Based on these equations, we derive an integration recursion relation and solve it iteratively. This approach enables us to explore the analytical structure of higher-pole reduction and provides a valuable tool for generating reduction coefficients efficiently., Comment: 30 pages

The Circular Electron Positron Collider (CEPC) has been proposed to enable more thorough and precise measurements of the properties of Higgs, W, and Z bosons, as well as to search for new physics. In response to the stringent performance requirements of the vertex detector for the CEPC, a baseline vertex detector prototype was tested and characterized for the first time using a 6 GeV electron beam at DESY II Test Beam Line 21. The baseline vertex detector prototype is designed with a cylindrical barrel structure that contains six double-sided detector modules (ladders). Each side of the ladder includes TaichuPix-3 sensors based on Monolithic Active Pixel Sensor (MAPS) technology, a flexible printed circuit, and a carbon fiber support structure. Additionally, the readout electronics and the Data Acquisition system were also examined during this beam test. The performance of the prototype was evaluated using an electron beam that passed through six ladders in a perpendicular direction. The offline data analysis indicates a spatial resolution of about 5 um, with detection efficiency exceeding 99 % and an impact parameter resolution of about 5.1 um. These promising results from this baseline vertex detector prototype mark a significant step toward realizing the optimal vertex detector for the CEPC.

While neural networks have excelled in video action recognition tasks, their black-box nature often obscures the understanding of their decision-making processes. Recent approaches used inherently interpretable models to analyze video actions in a manner akin to human reasoning. These models, however, usually fall short in performance compared to their black-box counterparts. In this work, we present a new framework named Language-guided Interpretable Action Recognition framework (LaIAR). LaIAR leverages knowledge from language models to enhance both the recognition capabilities and the interpretability of video models. In essence, we redefine the problem of understanding video model decisions as a task of aligning video and language models. Using the logical reasoning captured by the language model, we steer the training of the video model. This integrated approach not only improves the video model's adaptability to different domains but also boosts its overall performance. Extensive experiments on two complex video action datasets, Charades & CAD-120, validates the improved performance and interpretability of our LaIAR framework. The code of LaIAR is available at https://github.com/NingWang2049/LaIAR., Comment: Accepted by CVPR 2024

In contemporary society, the effective utilization of public resources remains a subject of significant concern. A common issue arises from defectors seeking to obtain an excessive share of these resources for personal gain, potentially leading to resource depletion. To mitigate this tragedy and ensure sustainable development of resources, implementing mechanisms to either reward those who adhere to distribution rules or penalize those who do not, appears advantageous. We introduce two models: a tax-reward model and a tax-punishment model, to address this issue. Our analysis reveals that in the tax-reward model, the evolutionary trajectory of the system is influenced not only by the tax revenue collected but also by the natural growth rate of the resources. Conversely, the tax-punishment model exhibits distinct characteristics when compared to the tax-reward model, notably the potential for bistability. In such scenarios, the selection of initial conditions is critical, as it can determine the system's path. Furthermore, our study identifies instances where the system lacks stable points, exemplified by a limit cycle phenomenon, underscoring the complexity and dynamism inherent in managing public resources using these models., Comment: Accepted by Proceedings of the Royal Society B-Biological Sciences

With the rise of large language models (LLMs), recent works have leveraged LLMs to improve the performance of click-through rate (CTR) prediction. However, we argue that a critical obstacle remains in deploying LLMs for practical use: the efficiency of LLMs when processing long textual user behaviors. As user sequences grow longer, the current efficiency of LLMs is inadequate for training on billions of users and items. To break through the efficiency barrier of LLMs, we propose Behavior Aggregated Hierarchical Encoding (BAHE) to enhance the efficiency of LLM-based CTR modeling. Specifically, BAHE proposes a novel hierarchical architecture that decouples the encoding of user behaviors from inter-behavior interactions. Firstly, to prevent computational redundancy from repeated encoding of identical user behaviors, BAHE employs the LLM's pre-trained shallow layers to extract embeddings of the most granular, atomic user behaviors from extensive user sequences and stores them in the offline database. Subsequently, the deeper, trainable layers of the LLM facilitate intricate inter-behavior interactions, thereby generating comprehensive user embeddings. This separation allows the learning of high-level user representations to be independent of low-level behavior encoding, significantly reducing computational complexity. Finally, these refined user embeddings, in conjunction with correspondingly processed item embeddings, are incorporated into the CTR model to compute the CTR scores. Extensive experimental results show that BAHE reduces training time and memory by five times for CTR models using LLMs, especially with longer user sequences. BAHE has been deployed in a real-world system, allowing for daily updates of 50 million CTR data on 8 A100 GPUs, making LLMs practical for industrial CTR prediction., Comment: Accepted by the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR), 2024

We propose a method based on the Hilbert-Huang transform (HHT) to recover the high-energy waveform of low-frequency quasi-periodic oscillations (LFQPOs). Based on the method, we successfully obtain the modulation of the phase-folded light curve above 170 keV using the QPO phase reconstructed at lower energies in MAXI J1535-571 with Insight-HXMT observations. A comprehensive simulation study is conducted to demonstrate that such modulation indeed originates from the QPO. Thus the highest energies turn out to significantly exceed the upper limit of ~100 keV for QPOs reported previously using the Fourier method, marking the first opportunity to study QPO properties above 100 keV in this source. Detailed analyses of these high-energy QPO profiles reveal different QPO properties between the 30-100 keV and 100-200 keV energy ranges: the phase lag remains relatively stable, and the amplitude slightly increases below ~100 keV, whereas above this threshold, soft phase lags and a decrease in amplitude are observed. Given the reports of a hard tail detection in broad spectroscopy, we propose that the newly discovered QPO properties above 100 keV are dominated by the hard tail component, possibly stemming from a relativistic jet. Our findings also indicate a strong correlation between the QPOs originating from the jet and corona, supporting the scenario of jet-corona coupling precssion. We emphasize that our proposed HHT-based method can serve as an efficient manner in expanding the high energy band for studying QPOs, thereby enhancing our understanding of their origin., Comment: 21 pages, 15 figures, accepted for publication in ApJL

We present the results obtained from an X-ray timing study of the new black hole candidate (BHC) Swift J1727.8-1613. The work is based on Hard X-ray Modulation Telescope (Insight-HXMT) observations carried out during the 2023 outburst. Prominent type-C low-frequency Quasi-periodic Oscillations (LFQPOs) are detected throughout the observations. With the substantial effective area of the Insight-HXMT at high energies, we examine the energy dependence of various parameters, including the centroid frequency, fractional rms, and phase lags of the type-C QPOs. Our findings align closely with those observed in high-inclination systems. During the initial stage of the outburst, a peaked noise component is also detected, the frequency of which is highly correlated with the LFQPO frequency, aligning with the Psaltis-Belloni-van der Klis (PBK) relation. By assuming that the peaked noise originates from the precession of the accretion disc, the spin of this source can be constrained. Our results suggest that this source may possess a high spin.

Structure information is critical for understanding the semantics of text-rich images, such as documents, tables, and charts. Existing Multimodal Large Language Models (MLLMs) for Visual Document Understanding are equipped with text recognition ability but lack general structure understanding abilities for text-rich document images. In this work, we emphasize the importance of structure information in Visual Document Understanding and propose the Unified Structure Learning to boost the performance of MLLMs. Our Unified Structure Learning comprises structure-aware parsing tasks and multi-grained text localization tasks across 5 domains: document, webpage, table, chart, and natural image. To better encode structure information, we design a simple and effective vision-to-text module H-Reducer, which can not only maintain the layout information but also reduce the length of visual features by merging horizontal adjacent patches through convolution, enabling the LLM to understand high-resolution images more efficiently. Furthermore, by constructing structure-aware text sequences and multi-grained pairs of texts and bounding boxes for publicly available text-rich images, we build a comprehensive training set DocStruct4M to support structure learning. Finally, we construct a small but high-quality reasoning tuning dataset DocReason25K to trigger the detailed explanation ability in the document domain. Our model DocOwl 1.5 achieves state-of-the-art performance on 10 visual document understanding benchmarks, improving the SOTA performance of MLLMs with a 7B LLM by more than 10 points in 5/10 benchmarks. Our codes, models, and datasets are publicly available at https://github.com/X-PLUG/mPLUG-DocOwl/tree/main/DocOwl1.5., Comment: 21 pages, 15 figures

Constrained variational inequality problems are recognized for their broad applications across various fields including machine learning and operations research. First-order methods have emerged as the standard approach for solving these problems due to their simplicity and scalability. However, they typically rely on projection or linear minimization oracles to navigate the feasible set, which becomes computationally expensive in practical scenarios featuring multiple functional constraints. Existing efforts to tackle such functional constrained variational inequality problems have centered on primal-dual algorithms grounded in the Lagrangian function. These algorithms along with their theoretical analysis often require the existence and prior knowledge of the optimal Lagrange multipliers. In this work, we propose a simple primal method, termed Constrained Gradient Method (CGM), for addressing functional constrained variational inequality problems, without necessitating any information on the optimal Lagrange multipliers. We establish a non-asymptotic convergence analysis of the algorithm for variational inequality problems with monotone operators under smooth constraints. Remarkably, our algorithms match the complexity of projection-based methods in terms of operator queries for both monotone and strongly monotone settings, while utilizing significantly cheaper oracles based on quadratic programming. Furthermore, we provide several numerical examples to evaluate the efficacy of our algorithms.

Reinforcement Learning from AI Feedback (RLAIF) has the advantages of shorter annotation cycles and lower costs over Reinforcement Learning from Human Feedback (RLHF), making it highly efficient during the rapid strategy iteration periods of large language model (LLM) training. Using ChatGPT as a labeler to provide feedback on open-domain prompts in RLAIF training, we observe an increase in human evaluators' preference win ratio for model responses, but a decrease in evaluators' satisfaction rate. Analysis suggests that the decrease in satisfaction rate is mainly due to some responses becoming less helpful, particularly in terms of correctness and truthfulness, highlighting practical limitations of basic RLAIF. In this paper, we propose Hybrid Reinforcement Learning from AI Feedback (HRLAIF). This method enhances the accuracy of AI annotations for responses, making the model's helpfulness more robust in training process. Additionally, it employs AI for Red Teaming, further improving the model's harmlessness. Human evaluation results show that HRLAIF inherits the ability of RLAIF to enhance human preference for outcomes at a low cost while also improving the satisfaction rate of responses. Compared to the policy model before Reinforcement Learning (RL), it achieves an increase of 2.08\% in satisfaction rate, effectively addressing the issue of a decrease of 4.58\% in satisfaction rate after basic RLAIF., Comment: 18 pages, 7 figures

Intelligent Tutoring Systems (ITSs) have significantly enhanced adult literacy training, a key factor for societal participation, employment opportunities, and lifelong learning. Our study investigates the application of advanced AI models, including Large Language Models (LLMs) like GPT-4, for predicting learning performance in adult literacy programs in ITSs. This research is motivated by the potential of LLMs to predict learning performance based on its inherent reasoning and computational capabilities. By using reading comprehension datasets from the ITS, AutoTutor, we evaluate the predictive capabilities of GPT-4 versus traditional machine learning methods in predicting learning performance through five-fold cross-validation techniques. Our findings show that the GPT-4 presents the competitive predictive abilities with traditional machine learning methods such as Bayesian Knowledge Tracing, Performance Factor Analysis, Sparse Factor Analysis Lite (SPARFA-Lite), tensor factorization and eXtreme Gradient Boosting (XGBoost). While XGBoost (trained on local machine) outperforms GPT-4 in predictive accuracy, GPT-4-selected XGBoost and its subsequent tuning on the GPT-4 platform demonstrates superior performance compared to local machine execution. Moreover, our investigation into hyper-parameter tuning by GPT-4 versus grid-search suggests comparable performance, albeit with less stability in the automated approach, using XGBoost as the case study. Our study contributes to the field by highlighting the potential of integrating LLMs with traditional machine learning models to enhance predictive accuracy and personalize adult literacy education, setting a foundation for future research in applying LLMs within ITSs., Comment: 26TH International Conference on Human-Computer Interaction

Kinematics restrict the ability of rare charm decays to explore the charged Lepton Flavor Violation processes mediated by the quark-level $c\to u \ell \tau$ transition. To fill the gap, we propose exploring new physics (NP) through the quasielastic scattering process $e^-p\to \tau^-\Lambda_c$ and the polarization of the $\tau$ lepton. As analyzing modes for the $\tau$ polarization, we consider the decays $\tau^-\to \pi^-\nu_{\tau}$, $\tau^-\to \rho^-\nu_{\tau}$, and $\tau^- \to \ell^-\bar{\nu}_{\ell}\nu_{\tau}$, and show that the $\tau$ polarization components can be extracted from analyzing the kinematics of the $\tau$ visible decay products. In the framework of a general low-energy effective Lagrangian, we then perform a detailed analysis of the polarization components in various aspects and scrutinize possible NP signals. With one upcoming experimental setup, we finally demonstrate promising event rate can be expected for the cascade process and, even in the worst-case scenario -- no signals is observed at all -- it can still provide a competitive potential for constraining the NP, compared with those from the high-$p_T$ dilepton invariant mass tails at high-energy colliders., Comment: 20 pages, 4 figures, 5 tables

Large Multimodal Models (LMMs) often suffer from multimodal hallucinations, wherein they may create content that is not present in the visual inputs. In this paper, we explore a new angle of this issue: overly detailed training data hinders the model's ability to timely terminate generation, leading to continued outputs beyond visual perception limits. By investigating how the model decides to terminate generation with EOS, the special end-of-sentence token, we find that the model assesses the completeness of the entire sequence by comparing the generated text with the image. This observation suggests that the model possesses an inherent potential of making proper EOS decisions based on its visual perception to avoid overly lengthy outputs. To take advantage of such potential, we explore two methods to mitigate multimodal hallucinations: a training objective that enables the model to reduce hallucinations by learning from regular instruction data, and a data filtering strategy to prevent harmful training data from exacerbating model hallucinations. Both methods significantly improve the hallucination performance of LMMs, without requiring any additional data or knowledge., Comment: Accepted to ACL 2024

XML configurations are integral to the Android development framework, particularly in the realm of UI display. However, these configurations can introduce compatibility issues (bugs), resulting in divergent visual outcomes and system crashes across various Android API versions (levels). In this study, we systematically investigate LLM-based approaches for detecting and repairing configuration compatibility bugs. Our findings highlight certain limitations of LLMs in effectively identifying and resolving these bugs, while also revealing their potential in addressing complex, hard-to-repair issues that traditional tools struggle with. Leveraging these insights, we introduce the LLM-CompDroid framework, which combines the strengths of LLMs and traditional tools for bug resolution. Our experimental results demonstrate a significant enhancement in bug resolution performance by LLM-CompDroid, with LLM-CompDroid-GPT-3.5 and LLM-CompDroid-GPT-4 surpassing the state-of-the-art tool, ConfFix, by at least 9.8% and 10.4% in both Correct and Correct@k metrics, respectively. This innovative approach holds promise for advancing the reliability and robustness of Android applications, making a valuable contribution to the field of software development.

Graph contrastive learning (GCL) has emerged as a pivotal technique in the domain of graph representation learning. A crucial aspect of effective GCL is the caliber of generated positive and negative samples, which is intrinsically dictated by their resemblance to the original data. Nevertheless, precise control over similarity during sample generation presents a formidable challenge, often impeding the effective discovery of representative graph patterns. To address this challenge, we propose an innovative framework: Adversarial Curriculum Graph Contrastive Learning (ACGCL), which capitalizes on the merits of pair-wise augmentation to engender graph-level positive and negative samples with controllable similarity, alongside subgraph contrastive learning to discern effective graph patterns therein. Within the ACGCL framework, we have devised a novel adversarial curriculum training methodology that facilitates progressive learning by sequentially increasing the difficulty of distinguishing the generated samples. Notably, this approach transcends the prevalent sparsity issue inherent in conventional curriculum learning strategies by adaptively concentrating on more challenging training data. Finally, a comprehensive assessment of ACGCL is conducted through extensive experiments on six well-known benchmark datasets, wherein ACGCL conspicuously surpasses a set of state-of-the-art baselines.

The potential of Machine Learning Control (MLC) in HVAC systems is hindered by its opaque nature and inference mechanisms, which is challenging for users and modelers to fully comprehend, ultimately leading to a lack of trust in MLC-based decision-making. To address this challenge, this paper investigates and explores Interpretable Machine Learning (IML), a branch of Machine Learning (ML) that enhances transparency and understanding of models and their inferences, to improve the credibility of MLC and its industrial application in HVAC systems. Specifically, we developed an innovative framework that combines the principles of Shapley values and the in-context learning feature of Large Language Models (LLMs). While the Shapley values are instrumental in dissecting the contributions of various features in ML models, LLM provides an in-depth understanding of rule-based parts in MLC; combining them, LLM further packages these insights into a coherent, human-understandable narrative. The paper presents a case study to demonstrate the feasibility of the developed IML framework for model predictive control-based precooling under demand response events in a virtual testbed. The results indicate that the developed framework generates and explains the control signals in accordance with the rule-based rationale.

The rapid progression in artificial intelligence has facilitated the emergence of large language models like ChatGPT, offering potential applications extending into specialized engineering modeling, especially physics-based building energy modeling. This paper investigates the innovative integration of large language models with building energy modeling software, focusing specifically on the fusion of ChatGPT with EnergyPlus. A literature review is first conducted to reveal a growing trend of incorporating of large language models in engineering modeling, albeit limited research on their application in building energy modeling. We underscore the potential of large language models in addressing building energy modeling challenges and outline potential applications including 1) simulation input generation, 2) simulation output analysis and visualization, 3) conducting error analysis, 4) co-simulation, 5) simulation knowledge extraction and training, and 6) simulation optimization. Three case studies reveal the transformative potential of large language models in automating and optimizing building energy modeling tasks, underscoring the pivotal role of artificial intelligence in advancing sustainable building practices and energy efficiency. The case studies demonstrate that selecting the right large language model techniques is essential to enhance performance and reduce engineering efforts. Besides direct use of large language models, three specific techniques were utilized: 1) prompt engineering, 2) retrieval-augmented generation, and 3) multi-agent large language models. The findings advocate a multidisciplinary approach in future artificial intelligence research, with implications extending beyond building energy modeling to other specialized engineering modeling.

Accurately modeling the protein fitness landscapes holds great importance for protein engineering. Recently, due to their capacity and representation ability, pre-trained protein language models have achieved state-of-the-art performance in predicting protein fitness without experimental data. However, their predictions are limited in accuracy as well as interpretability. Furthermore, such deep learning models require abundant labeled training examples for performance improvements, posing a practical barrier. In this work, we introduce FSFP, a training strategy that can effectively optimize protein language models under extreme data scarcity. By combining the techniques of meta-transfer learning, learning to rank, and parameter-efficient fine-tuning, FSFP can significantly boost the performance of various protein language models using merely tens of labeled single-site mutants from the target protein. The experiments across 87 deep mutational scanning datasets underscore its superiority over both unsupervised and supervised approaches, revealing its potential in facilitating AI-guided protein design.