Search

Position bias has proven to be a prevalent issue of modern language models (LMs), where the models prioritize content based on its position within the given context. This bias often leads to unexpected model failures and hurts performance, robustness, and reliability across various applications. Our mechanistic analysis attributes the position bias to two components employed in nearly all state-of-the-art LMs: causal attention and relative positional encodings. Specifically, we find that causal attention generally causes models to favor distant content, while relative positional encodings like RoPE prefer nearby ones based on the analysis of retrieval-augmented question answering (QA). Further, our empirical study on object detection reveals that position bias is also present in vision-language models (VLMs). Based on the above analyses, we propose to ELIMINATE position bias caused by different input segment orders (e.g., options in LM-as-a-judge, retrieved documents in QA) in a TRAINING-FREE ZERO-SHOT manner. Our method changes the causal attention to bidirectional attention between segments and utilizes model attention values to decide the relative orders of segments instead of using the order provided in input prompts, therefore enabling Position-INvariant inferencE (PINE) at the segment level. By eliminating position bias, models achieve better performance and reliability in downstream tasks where position bias widely exists, such as LM-as-a-judge and retrieval-augmented QA. Notably, PINE is especially useful when adapting LMs for evaluating reasoning pairs: it consistently provides 8 to 10 percentage points performance gains in most cases, and makes Llama-3-70B-Instruct perform even better than GPT-4-0125-preview on the RewardBench reasoning subset., Comment: 18 pages, 5 figures

GPT-4 is often heralded as a leading commercial AI offering, sparking debates over its potential as a steppingstone toward artificial general intelligence. But does it possess consciousness? This paper investigates this key question using the nine qualitative measurements of the Building Blocks theory. GPT-4's design, architecture and implementation are compared to each of the building blocks of consciousness to determine whether it has achieved the requisite milestones to be classified as conscious or, if not, how close to consciousness GPT-4 is. Our assessment is that, while GPT-4 in its native configuration is not currently conscious, current technological research and development is sufficient to modify GPT-4 to have all the building blocks of consciousness. Consequently, we argue that the emergence of a conscious AI model is plausible in the near term. The paper concludes with a comprehensive discussion of the ethical implications and societal ramifications of engineering conscious AI entities., Comment: Accepted for publication in the Journal of AI Consciousness

Recent years have witnessed significant advancements in light field image super-resolution (LFSR) owing to the progress of modern neural networks. However, these methods often face challenges in capturing long-range dependencies (CNN-based) or encounter quadratic computational complexities (Transformer-based), which limit their performance. Recently, the State Space Model (SSM) with selective scanning mechanism (S6), exemplified by Mamba, has emerged as a superior alternative in various vision tasks compared to traditional CNN- and Transformer-based approaches, benefiting from its effective long-range sequence modeling capability and linear-time complexity. Therefore, integrating S6 into LFSR becomes compelling, especially considering the vast data volume of 4D light fields. However, the primary challenge lies in \emph{designing an appropriate scanning method for 4D light fields that effectively models light field features}. To tackle this, we employ SSMs on the informative 2D slices of 4D LFs to fully explore spatial contextual information, complementary angular information, and structure information. To achieve this, we carefully devise a basic SSM block characterized by an efficient SS2D mechanism that facilitates more effective and efficient feature learning on these 2D slices. Based on the above two designs, we further introduce an SSM-based network for LFSR termed LFMamba. Experimental results on LF benchmarks demonstrate the superior performance of LFMamba. Furthermore, extensive ablation studies are conducted to validate the efficacy and generalization ability of our proposed method. We expect that our LFMamba shed light on effective representation learning of LFs with state space models.

Federated multi-view clustering offers the potential to develop a global clustering model using data distributed across multiple devices. However, current methods face challenges due to the absence of label information and the paramount importance of data privacy. A significant issue is the feature heterogeneity across multi-view data, which complicates the effective mining of complementary clustering information. Additionally, the inherent incompleteness of multi-view data in a distributed setting can further complicate the clustering process. To address these challenges, we introduce a federated incomplete multi-view clustering framework with heterogeneous graph neural networks (FIM-GNNs). In the proposed FIM-GNNs, autoencoders built on heterogeneous graph neural network models are employed for feature extraction of multi-view data at each client site. At the server level, heterogeneous features from overlapping samples of each client are aggregated into a global feature representation. Global pseudo-labels are generated at the server to enhance the handling of incomplete view data, where these labels serve as a guide for integrating and refining the clustering process across different data views. Comprehensive experiments have been conducted on public benchmark datasets to verify the performance of the proposed FIM-GNNs in comparison with state-of-the-art algorithms.

Localization is a critical technology for various applications ranging from navigation and surveillance to assisted living. Localization systems typically fuse information from sensors viewing the scene from different perspectives to estimate the target location while also employing multiple modalities for enhanced robustness and accuracy. Recently, such systems have employed end-to-end deep neural models trained on large datasets due to their superior performance and ability to handle data from diverse sensor modalities. However, such neural models are often trained on data collected from a particular set of sensor poses (i.e., locations and orientations). During real-world deployments, slight deviations from these sensor poses can result in extreme inaccuracies. To address this challenge, we introduce FlexLoc, which employs conditional neural networks to inject node perspective information to adapt the localization pipeline. Specifically, a small subset of model weights are derived from node poses at run time, enabling accurate generalization to unseen perspectives with minimal additional overhead. Our evaluations on a multimodal, multiview indoor tracking dataset showcase that FlexLoc improves the localization accuracy by almost 50% in the zero-shot case (no calibration data available) compared to the baselines. The source code of FlexLoc is available at https://github.com/nesl/FlexLoc.

We introduce linear regression using physics-based basis functions optimized through the geometry of an inner product space. This method addresses the challenge of surrogate modeling with high-dimensional input, as the physics-based basis functions encode problem-specific information. We demonstrate the method using a proof-of-concept nonlinear random vibration example.

The open-source community is experiencing a surge in the release of large language models (LLMs) that are trained to follow instructions and align with human preference. However, further training to improve them still requires expensive computational resources and data annotations. Is it possible to bypass additional training and cost-effectively acquire better-aligned models? Inspired by the literature on model interpolation, we propose a simple method called ExPO to boost LLMs' alignment with human preference. Utilizing a model that has undergone alignment training (e.g., via DPO or RLHF) and its initial SFT checkpoint, ExPO directly obtains a better-aligned model by extrapolating from the weights of the initial and the aligned models, which implicitly optimizes the alignment objective via first-order approximation. Through experiments with twelve open-source LLMs on HuggingFace, we demonstrate that ExPO consistently improves off-the-shelf DPO/RLHF models, as evaluated on the mainstream LLM benchmarks AlpacaEval 2.0 and MT-Bench. Moreover, ExPO exhibits remarkable scalability across various model sizes (from 1.8B to 70B) and capabilities. Through controlled experiments and further empirical analyses, we shed light on the essence of ExPO amplifying the reward signal learned during alignment training. Our work demonstrates the efficacy of model extrapolation in expediting the alignment of LLMs with human preference, suggesting a promising direction for future research., Comment: Add theoretical explanation and more evaluation results

This survey comprehensively reviews the multi-dimensionality of game scenario diversity, spotlighting the innovative use of procedural content generation and other fields as cornerstones for enriching player experiences through diverse game scenarios. By traversing a wide array of disciplines, from affective modeling and multi-agent systems to psychological studies, our research underscores the importance of diverse game scenarios in gameplay and education. Through a taxonomy of diversity metrics and evaluation methods, we aim to bridge the current gaps in literature and practice, offering insights into effective strategies for measuring and integrating diversity in game scenarios. Our analysis highlights the necessity for a unified taxonomy to aid developers and researchers in crafting more engaging and varied game worlds. This survey not only charts a path for future research in diverse game scenarios but also serves as a handbook for industry practitioners seeking to leverage diversity as a key component of game design and development.

The active learning (AL) technique, one of the state-of-the-art methods for constructing surrogate models, has shown high accuracy and efficiency in forward uncertainty quantification (UQ) analysis. This paper provides a comprehensive study on AL-based global surrogates for computing the full distribution function, i.e., the cumulative distribution function (CDF) and the complementary CDF (CCDF). To this end, we investigate the three essential components for building surrogates, i.e., types of surrogate models, enrichment methods for experimental designs, and stopping criteria. For each component, we choose several representative methods and study their desirable configurations. In addition, we devise a uniform design (i.e., space-filling design) as a baseline for measuring the improvement of using AL. Combining all the representative methods, a total of 1,920 UQ analyses are carried out to solve 16 benchmark examples. The performance of the selected strategies is evaluated based on accuracy and efficiency. In the context of full distribution estimation, this study concludes that (i) AL techniques cannot provide a systematic improvement compared with uniform designs, (ii) the recommended surrogate modeling methods depend on the features of the problems (especially the local nonlinearity), target accuracy, and computational budget.

Automatic test generation plays a critical role in software quality assurance. While the recent advances in Search-Based Software Testing (SBST) and Large Language Models (LLMs) have shown promise in generating useful tests, these techniques still struggle to cover certain branches. Reaching these hard-to-cover branches usually requires constructing complex objects and resolving intricate inter-procedural dependencies in branch conditions, which poses significant challenges for existing test generation techniques. In this work, we propose TELPA, a novel technique aimed at addressing these challenges. Its key insight lies in extracting real usage scenarios of the target method under test to learn how to construct complex objects and extracting methods entailing inter-procedural dependencies with hard-to-cover branches to learn the semantics of branch constraints. To enhance efficiency and effectiveness, TELPA identifies a set of ineffective tests as counter-examples for LLMs and employs a feedback-based process to iteratively refine these counter-examples. Then, TELPA integrates program analysis results and counter-examples into the prompt, guiding LLMs to gain deeper understandings of the semantics of the target method and generate diverse tests that can reach the hard-to-cover branches. Our experimental results on 27 open-source Python projects demonstrate that TELPA significantly outperforms the state-of-the-art SBST and LLM-based techniques, achieving an average improvement of 31.39% and 22.22% in terms of branch coverage., Comment: 11 pages, 4 figures

This study introduces the long-range Ising model from statistical mechanics to the Performance-Based Earthquake Engineering (PBEE) framework for regional seismic damage analysis. The application of the PBEE framework at a regional scale involves estimating the damage states of numerous structures, typically performed using fragility function-based stochastic simulations. However, these simulations often assume conditional independence or employ simplistic dependency models among the damage states of structures, leading to significant misrepresentation of regional risk. The Ising model addresses this issue by converting the available information on binary damage states (safe or failure) into a joint probability mass function, leveraging the principle of maximum entropy. The Ising model offers two main benefits: (1) it requires only the first- and second-order cross-moments, enabling seamless integration with the existing PBEE framework, and (2) it provides meaningful physical interpretations of the model parameters, facilitating the uncovering of insights not apparent from data. To demonstrate the proposed method, we applied the Ising model to 156 buildings in Antakya, Turkey, using post-hazard damage evaluation data, and to 182 buildings in Pacific Heights, San Francisco, using simulated data from the Regional Resilience Determination (R2D) tool. In both instances, the Ising model accurately reproduces the provided information and generates meaningful insights into regional damage. The study also investigates the change in Ising model parameters under varying earthquake magnitudes, along with the mean-field approximation, further facilitating the applicability of the proposed approach.

Small bubbles in fluids raise to the surface due to Archimede's force. Remarkably, in turbulent flows this process is severely hindered by the presence of vortex filaments, which act as moving potential wells, dynamically trapping light particles and bubbles. Quantifying the statistical weights and roles of vortex filaments in turbulence is, however, still an outstanding experimental and computational challenge due to their small scale, fast chaotic motion, and transient nature. Here we show that, under the influence of a modulated oscillatory forcing, the collective bubble behaviour switches from a dynamically localized (insulator) to a delocalized (conductor) state. Additionally, we find that by varying the forcing frequency and amplitude, a remarkable resonant phenomenon between light particles and small-scale vortex filamenents emerges, likening particle behavior to a forced damped oscillator. We discuss how these externally actuated bubbles can be used as a new type of microscopic probes to investigate the space-time statistical properties of the smallest turbulence scales, allowing to quantitatively measure physical characteristics of vortex filaments. We develop a superposition model that is in excellent agreement with the simulation data of the particle dynamics which reveals the fraction of localized/delocalized particles as well as characteristics of the potential landscape induced by vortices in turbulence. Our approach paves the way to innovative ways to accurately measure turbulent properties and to the possibility to control light particles and bubble motions in turbulence with potential applications to wave breaking, medical imaging, drug/gene delivery, chemical reactions, wastewater treatment, and industrial mixing., Comment: 10 pages, 5 figures

Digital twinning in structural engineering is a rapidly evolving technology that aims to eliminate the gap between physical systems and their digital models through real-time sensing, visualization, and control techniques. Although digital twins can offer dynamic insights into physical systems, their accuracy is inevitably compromised by uncertainties in sensing, modeling, simulation, and controlling. This paper proposes a specialized digital twin formulation, named Risk Twin, designed for real-time risk visualization and risk-informed control of structural systems. Integrating structural reliability and Bayesian inference methods with digital twining techniques, Risk Twin can analyze and visualize the reliability indices for structural components in real-time. To facilitate real-time inference and reliability updating, a "simulation-free" scheme is proposed, leveraging precomputed quantities prepared during an offline phase for rapid inference in the online phase. Proof-of-concept numerical and real-world Risk Twins are constructed to showcase the proposed concepts.

Constantly locating moving objects, i.e., geospatial tracking, is essential for autonomous building infrastructure. Accurate and robust geospatial tracking often leverages multimodal sensor fusion algorithms, which require large datasets with time-aligned, synchronized data from various sensor types. However, such datasets are not readily available. Hence, we propose GDTM, a nine-hour dataset for multimodal object tracking with distributed multimodal sensors and reconfigurable sensor node placements. Our dataset enables the exploration of several research problems, such as optimizing architectures for processing multimodal data, and investigating models' robustness to adverse sensing conditions and sensor placement variances. A GitHub repository containing the code, sample data, and checkpoints of this work is available at https://github.com/nesl/GDTM.

Federated learning (FL) is a promising framework for learning from distributed data while maintaining privacy. The development of efficient FL algorithms encounters various challenges, including heterogeneous data and systems, limited communication capacities, and constrained local computational resources. Recently developed FedADMM methods show great resilience to both data and system heterogeneity. However, they still suffer from performance deterioration if the hyperparameters are not carefully tuned. To address this issue, we propose an inexact and self-adaptive FedADMM algorithm, termed FedADMM-InSa. First, we design an inexactness criterion for the clients' local updates to eliminate the need for empirically setting the local training accuracy. This inexactness criterion can be assessed by each client independently based on its unique condition, thereby reducing the local computational cost and mitigating the undesirable straggle effect. The convergence of the resulting inexact ADMM is proved under the assumption of strongly convex loss functions. Additionally, we present a self-adaptive scheme that dynamically adjusts each client's penalty parameter, enhancing algorithm robustness by mitigating the need for empirical penalty parameter choices for each client. Extensive numerical experiments on both synthetic and real-world datasets are conducted. As validated by some numerical tests, our proposed algorithm can reduce the clients' local computational load significantly and also accelerate the learning process compared to the vanilla FedADMM.

In the current environment, psychological issues are prevalent and widespread, with social media serving as a key outlet for individuals to share their feelings. This results in the generation of vast quantities of data daily, where negative emotions have the potential to precipitate crisis situations. There is a recognized need for models capable of efficient analysis. While pre-trained language models have demonstrated their effectiveness broadly, there's a noticeable gap in pre-trained models tailored for specialized domains like psychology. To address this, we have collected a huge dataset from Chinese social media platforms and enriched it with publicly available datasets to create a comprehensive database encompassing 3.36 million text entries. To enhance the model's applicability to psychological text analysis, we integrated psychological lexicons into the pre-training masking mechanism. Building on an existing Chinese language model, we performed adaptive training to develop a model specialized for the psychological domain. We evaluated our model's performance across six public datasets, where it demonstrated improvements compared to eight other models. Additionally, in the qualitative comparison experiment, our model provided psychologically relevant predictions given the masked sentences. Due to concerns regarding data privacy, the dataset will not be made publicly available. However, we have made the pre-trained models and codes publicly accessible to the community via: https://github.com/zwzzzQAQ/Chinese-MentalBERT.

In this work, we consider a two-dimensional (2D) dilaton gravity model where the dilaton kinetic term $\mathcal{X}$ is modified by an additional derivative coupling term $\alpha\mathcal{X}^2$. In the case with a canonical scalar matter field, the field equations of this model have a simple first-order formalism, from which exact static kink solutions can be constructed. The novelty of these solutions is that the corresponding metric can be asymptotically flat rather than asymptotically anti de Sitter. The linear stability and the localization of scalar matter fields are also studied. It was found that the solutions are stable against small linear perturbations, and the localization of scalar matter fields can be realized by introducing scalar-kink interactions., Comment: 7 pages, 2 figures

We introduce a method to construct a stochastic surrogate model from the results of dimensionality reduction in forward uncertainty quantification. The hypothesis is that the high-dimensional input augmented by the output of a computational model admits a low-dimensional representation. This assumption can be met by numerous uncertainty quantification applications with physics-based computational models. The proposed approach differs from a sequential application of dimensionality reduction followed by surrogate modeling, as we "extract" a surrogate model from the results of dimensionality reduction in the input-output space. This feature becomes desirable when the input space is genuinely high-dimensional. The proposed method also diverges from the Probabilistic Learning on Manifold, as a reconstruction mapping from the feature space to the input-output space is circumvented. The final product of the proposed method is a stochastic simulator that propagates a deterministic input into a stochastic output, preserving the convenience of a sequential "dimensionality reduction + Gaussian process regression" approach while overcoming some of its limitations. The proposed method is demonstrated through two uncertainty quantification problems characterized by high-dimensional input uncertainties.

This paper studies the alignment process of generative models with Reinforcement Learning from Human Feedback (RLHF). We first identify the primary challenges of existing popular methods like offline PPO and offline DPO as lacking in strategical exploration of the environment. Then, to understand the mathematical principle of RLHF, we consider a standard mathematical formulation, the reverse-KL regularized contextual bandit for RLHF. Despite its widespread practical application, a rigorous theoretical analysis of this formulation remains open. We investigate its behavior in three distinct settings -- offline, online, and hybrid -- and propose efficient algorithms with finite-sample theoretical guarantees. Moving towards practical applications, our framework, with a robust approximation of the information-theoretical policy improvement oracle, naturally gives rise to several novel RLHF algorithms. This includes an iterative version of the Direct Preference Optimization (DPO) algorithm for online settings, and a multi-step rejection sampling strategy for offline scenarios. Our empirical evaluations on real-world alignment experiment of large language model demonstrate that these proposed methods significantly surpass existing strong baselines, such as DPO and Rejection Sampling Optimization (RSO), showcasing the connections between solid theoretical foundations and their potent practical implementations., Comment: 53 pages; theoretical study and algorithmic design of iterative RLHF and DPO

Hong Kong's anti-ELAB movement had a significant impact on the stock market the stock price of listed companies. Using the number of protestors as the measurement of daily protesting intensity from 2019/6/6 to 2020/1/17, this paper documents that the stock price of listed companies associated with the pan-democratic parties were more negatively affected by protesting than other companies. Furthermore, this paper finds that after the implementation of the anti-mask law, protesting had a positive impact on red chips but a negative impact on companies related to pan-democracy parties. Therefore, this paper believes that after the central government and the HKSAR government adopted strict measures to stop violence and chaos, the value of the political connection of red chips became positive while the value of the connection with pan-democracy parties became negative., Comment: 34 pages, 13 tables

Perturbed by natural hazards, community-level infrastructure networks operate like many-body systems, with behaviors emerging from coupling individual component dynamics with group correlations and interactions. It follows that we can borrow methods from statistical physics to study the response of infrastructure systems to natural disasters. This study aims to construct a joint probability distribution model to describe the post-hazard state of infrastructure networks and propose an efficient surrogate model of the joint distribution for large-scale systems. Specifically, we present maximum entropy modeling of the regional impact of natural hazards on civil infrastructures. Provided with the current state of knowledge, the principle of maximum entropy yields the ``most unbiased`` joint distribution model for the performances of infrastructures. In the general form, the model can handle multivariate performance states and higher-order correlations. In a particular yet typical scenario of binary performance state variables with knowledge of their mean and pairwise correlation, the joint distribution reduces to the Ising model in statistical physics. In this context, we propose using a dichotomized Gaussian model as an efficient surrogate for the maximum entropy model, facilitating the application to large systems. Using the proposed method, we investigate the seismic collective behavior of a large-scale road network (with 8,694 nodes and 26,964 links) in San Francisco, showcasing the non-trivial collective behaviors of infrastructure systems.

Various simulation-based and analytical methods have been developed to evaluate the seismic fragilities of individual structures. However, a community's seismic safety and resilience are substantially affected by network reliability, determined not only by component fragilities but also by network topology and commodity/information flows. However, seismic reliability analyses of networks often encounter significant challenges due to complex network topologies, interdependencies among ground motions, and low failure probabilities. This paper proposes to overcome these challenges by a variance-reduction method for network fragility analysis using subset simulation. The binary network limit-state function in the subset simulation is reformulated into more informative piecewise continuous functions. The proposed limit-state functions quantify the proximity of each sample to a potential network failure domain, thereby enabling the construction of specialized intermediate failure events, which can be utilized in subset simulation and other sequential Monte Carlo approaches. Moreover, by discovering an implicit connection between intermediate failure events and seismic intensity, we propose a technique to obtain the entire network fragility curve with a single execution of specialized subset simulation. Numerical examples demonstrate that the proposed method can effectively evaluate system-level fragility for large-scale networks.

Instruction tuning enhances the instruction following ability of large language models by finetuning with supervised instruction data. Previous work proposes in-context instruction tuning (ICIT) where specific positive or negative examples are incorporated into the prompt for better performance. In this work, we propose PACIT, a simple and effective in-context instruction tuning method, inspired by the pedagogical concept of desirable difficulty. The PACIT method unlocks the power of examples by encouraging the model to actively learn to grasp the distinctions between the positive and negative examples instead of merely reading. The model is expected to first verify the correctness of the provided example according to the task description, which is then set as the condition for generating a better response to the task instance. Our extensive experiments prove the effectiveness of PACIT, outperforming ICIT baseline on both in-domain and out-domain tasks up to 9.16 and 3.14 average ROUGE-L scores, respectively. Moreover, PACIT can notably enhance the performance of instruction tuning even when all positive and negative examples are generated with a self-instruct method., Comment: Accepted to ACL Findings 2024

Large Language Models (LLMs) have demonstrated remarkable capabilities in open-ended text generation tasks. However, the inherent open-ended nature of these tasks implies that there is always room for improvement in the quality of model responses. To address this challenge, various approaches have been proposed to enhance the performance of LLMs. There has been a growing focus on enabling LLMs to self-improve their response quality, thereby reducing the reliance on extensive human annotation efforts for collecting diverse and high-quality training data. Recently, prompting-based methods have been widely explored among self-improvement methods owing to their effectiveness, efficiency, and convenience. However, those methods usually require explicitly and thoroughly written rubrics as inputs to LLMs. It is expensive and challenging to manually derive and provide all necessary rubrics with a real-world complex goal for improvement (e.g., being more helpful and less harmful). To this end, we propose an ImPlicit Self-ImprovemenT (PIT) framework that implicitly learns the improvement goal from human preference data. PIT only requires preference data that are used to train reward models without extra human efforts. Specifically, we reformulate the training objective of reinforcement learning from human feedback (RLHF) -- instead of maximizing response quality for a given input, we maximize the quality gap of the response conditioned on a reference response. In this way, PIT is implicitly trained with the improvement goal of better aligning with human preferences. Experiments on two real-world datasets and one synthetic dataset show that our method significantly outperforms prompting-based methods., Comment: 28 pages, 5 figures, 4 tables

Aligning large language models (LLMs) with human preferences is essential for safe and useful LLMs. Previous works mainly adopt reinforcement learning (RLHF) and direct preference optimization (DPO) with human feedback for alignment. Nevertheless, they have certain drawbacks. One such limitation is that they can only align models with one preference at the training time (e.g., they cannot learn to generate concise responses when the preference data prefers detailed responses), or have certain constraints for the data format (e.g., DPO only supports pairwise preference data). To this end, prior works incorporate controllable generations for alignment to make language models learn multiple preferences and provide outputs with different preferences during inference if asked. Controllable generation also offers more flexibility with regard to data format (e.g., it supports pointwise preference data). Specifically, it uses different control tokens for different preferences during training and inference, making LLMs behave differently when required. Current controllable generation methods either use a special token or hand-crafted prompts as control tokens, and optimize them together with LLMs. As control tokens are typically much lighter than LLMs, this optimization strategy may not effectively optimize control tokens. To this end, we first use parameter-efficient tuning (e.g., prompting tuning and low-rank adaptation) to optimize control tokens and then fine-tune models for controllable generations, similar to prior works. Our approach, alignMEnt with parameter-Efficient Tuning (MEET), improves the quality of control tokens, thus improving controllable generation quality consistently by an apparent margin on two well-recognized datasets compared with prior works., Comment: 21 pages, 11 figures, 5 tables

This paper presents a physics and data co-driven surrogate modeling method for efficient rare event simulation of civil and mechanical systems with high-dimensional input uncertainties. The method fuses interpretable low-fidelity physical models with data-driven error corrections. The hypothesis is that a well-designed and well-trained simplified physical model can preserve salient features of the original model, while data-fitting techniques can fill the remaining gaps between the surrogate and original model predictions. The coupled physics-data-driven surrogate model is adaptively trained using active learning, aiming to achieve a high correlation and small bias between the surrogate and original model responses in the critical parametric region of a rare event. A final importance sampling step is introduced to correct the surrogate model-based probability estimations. Static and dynamic problems with input uncertainties modeled by random field and stochastic process are studied to demonstrate the proposed method.

The emerging field of artificial intelligence of things (AIoT, AI+IoT) is driven by the widespread use of intelligent infrastructures and the impressive success of deep learning (DL). With the deployment of DL on various intelligent infrastructures featuring rich sensors and weak DL computing capabilities, a diverse range of AIoT applications has become possible. However, DL models are notoriously resource-intensive. Existing research strives to realize near-/realtime inference of AIoT live data and low-cost training using AIoT datasets on resource-scare infrastructures. Accordingly, the accuracy and responsiveness of DL models are bounded by resource availability. To this end, the algorithm-system co-design that jointly optimizes the resource-friendly DL models and model-adaptive system scheduling improves the runtime resource availability and thus pushes the performance boundary set by the standalone level. Unlike previous surveys on resource-friendly DL models or hand-crafted DL compilers/frameworks with partially fine-tuned components, this survey aims to provide a broader optimization space for more free resource-performance tradeoffs. The cross-level optimization landscape involves various granularity, including the DL model, computation graph, operator, memory schedule, and hardware instructor in both on-device and distributed paradigms. Furthermore, due to the dynamic nature of AIoT context, which includes heterogeneous hardware, agnostic sensing data, varying user-specified performance demands, and resource constraints, this survey explores the context-aware inter-/intra-device controllers for automatic cross-level adaptation. Additionally, we identify some potential directions for resource-efficient AIoT systems. By consolidating problems and techniques scattered over diverse levels, we aim to help readers understand their connections and stimulate further discussions.

Federated Learning (FL) is a distributed learning paradigm that enables multiple clients to collaborate on building a machine learning model without sharing their private data. Although FL is considered privacy-preserved by design, recent data reconstruction attacks demonstrate that an attacker can recover clients' training data based on the parameters shared in FL. However, most existing methods fail to attack the most widely used horizontal Federated Averaging (FedAvg) scenario, where clients share model parameters after multiple local training steps. To tackle this issue, we propose an interpolation-based approximation method, which makes attacking FedAvg scenarios feasible by generating the intermediate model updates of the clients' local training processes. Then, we design a layer-wise weighted loss function to improve the data quality of reconstruction. We assign different weights to model updates in different layers concerning the neural network structure, with the weights tuned by Bayesian optimization. Finally, experimental results validate the superiority of our proposed approximate and weighted attack (AWA) method over the other state-of-the-art methods, as demonstrated by the substantial improvement in different evaluation metrics for image data reconstructions.

The demand for psychological counselling has grown significantly in recent years, particularly with the global outbreak of COVID-19, which has heightened the need for timely and professional mental health support. Online psychological counselling has emerged as the predominant mode of providing services in response to this demand. In this study, we propose the Psy-LLM framework, an AI-based assistive tool leveraging Large Language Models (LLMs) for question-answering in psychological consultation settings to ease the demand for mental health professions. Our framework combines pre-trained LLMs with real-world professional Q\&A from psychologists and extensively crawled psychological articles. The Psy-LLM framework serves as a front-end tool for healthcare professionals, allowing them to provide immediate responses and mindfulness activities to alleviate patient stress. Additionally, it functions as a screening tool to identify urgent cases requiring further assistance. We evaluated the framework using intrinsic metrics, such as perplexity, and extrinsic evaluation metrics, with human participant assessments of response helpfulness, fluency, relevance, and logic. The results demonstrate the effectiveness of the Psy-LLM framework in generating coherent and relevant answers to psychological questions. This article discusses the potential and limitations of using large language models to enhance mental health support through AI technologies.

In this study, the first observation of high-frequency instabilities driven by runaway electrons has been reported in the EXL-50 spherical torus using a high-frequency magnetic pickup coil. The central frequency of these instabilities is found to be exponentially dependent on the plasma density, similar to the dispersion relation of the whistler wave. The instability frequency displays chirping characteristics consistent with the Berk-Breizman model of beam instability. Theoretically, the excitation threshold of the instability driven by runaway electrons is related to the ratio of the runaway electron density to the background plasma density, and such a relationship is first demonstrated experimentally in this study. The instability can be stabilized by increasing the plasma density, consistent with the wave-particle resonance mechanism. This investigation demonstrates the controlled excitation of chirping instabilities in a tokamak plasma and reveals new features of these instabilities, thereby advancing the understanding of the mechanisms for controlling and mitigating runaway electrons.

Electromagnetic modes in the frequency range of 30-120MHz were observed in electron cyclotron wave (ECW) steady state plasmas on the ENN XuanLong-50 (EXL-50) spherical torus. These modes were found to have multiple bands of frequencies proportional to the Alfv\'en velocity. This indicates that the observed mode frequencies satisfy the dispersion relation of whistler waves. In addition, suppression of the whistler waves by the synergistic effect of Lower Hybrid Wave (LHW) and ECW was also observed. This suggests that the whistler waves were driven by temperature anisotropy of energetic electrons. These are the first such observations (not runaway discharge) made in magnetically confined toroidal plasmas and may have important implications for studying wave-particle interactions, RF wave current driver, and runaway electron control in future fusion devices.

Few-shot learning (FSL) is one of the significant and hard problems in the field of image classification. However, in contrast to the rapid development of the visible light dataset, the progress in SAR target image classification is much slower. The lack of unified benchmark is a key reason for this phenomenon, which may be severely overlooked by the current literature. The researchers of SAR target image classification always report their new results on their own datasets and experimental setup. It leads to inefficiency in result comparison and impedes the further progress of this area. Motivated by this observation, we propose a novel few-shot SAR image classification benchmark (FewSAR) to address this issue. FewSAR consists of an open-source Python code library of 15 classic methods in three categories for few-shot SAR image classification. It provides an accessible and customizable testbed for different few-shot SAR image classification task. To further understanding the performance of different few-shot methods, we establish evaluation protocols and conduct extensive experiments within the benchmark. By analyzing the quantitative results and runtime under the same setting, we observe that the accuracy of metric learning methods can achieve the best results. Meta-learning methods and fine-tuning methods perform poorly on few-shot SAR images, which is primarily due to the bias of existing datasets. We believe that FewSAR will open up a new avenue for future research and development, on real-world challenges at the intersection of SAR image classification and few-shot deep learning. We will provide our code for the proposed FewSAR at https://github.com/solarlee/FewSAR., Comment: 7 pages, 4 figures

Visual object tracking has seen significant progress in recent years. However, the vast majority of this work focuses on tracking objects within the image plane of a single camera and ignores the uncertainty associated with predicted object locations. In this work, we focus on the geospatial object tracking problem using data from a distributed camera network. The goal is to predict an object's track in geospatial coordinates along with uncertainty over the object's location while respecting communication constraints that prohibit centralizing raw image data. We present a novel single-object geospatial tracking data set that includes high-accuracy ground truth object locations and video data from a network of four cameras. We present a modeling framework for addressing this task including a novel backbone model and explore how uncertainty calibration and fine-tuning through a differentiable tracker affect performance.

Large language models (LLMs) have initiated a paradigm shift in transfer learning. In contrast to the classic pretraining-then-finetuning procedure, in order to use LLMs for downstream prediction tasks, one only needs to provide a few demonstrations, known as in-context examples, without adding more or updating existing model parameters. This in-context learning (ICL) capability of LLMs is intriguing, and it is not yet fully understood how pretrained LLMs acquire such capabilities. In this paper, we investigate the reason why a transformer-based language model can accomplish in-context learning after pre-training on a general language corpus by proposing one hypothesis that LLMs can simulate kernel regression with internal representations when faced with in-context examples. More concretely, we first prove that Bayesian inference on in-context prompts can be asymptotically understood as kernel regression $\hat y = \sum_i y_i K(x, x_i)/\sum_i K(x, x_i)$ as the number of in-context demonstrations grows. Then, we empirically investigate the in-context behaviors of language models. We find that during ICL, the attention and hidden features in LLMs match the behaviors of a kernel regression. Finally, our theory provides insights into multiple phenomena observed in the ICL field: why retrieving demonstrative samples similar to test samples can help, why ICL performance is sensitive to the output formats, and why ICL accuracy benefits from selecting in-distribution and representative samples., Comment: 9 pages, 4 figures

Knowledge distillation (KD) requires sufficient data to transfer knowledge from large-scale teacher models to small-scale student models. Therefore, data augmentation has been widely used to mitigate the shortage of data under specific scenarios. Classic data augmentation techniques, such as synonym replacement and k-nearest-neighbors, are initially designed for fine-tuning. To avoid severe semantic shifts and preserve task-specific labels, those methods prefer to change only a small proportion of tokens (e.g., changing 10% tokens is generally the best option for fine-tuning). However, such data augmentation methods are sub-optimal for knowledge distillation since the teacher model could provide label distributions and is more tolerant to semantic shifts. We first observe that KD prefers as much data as possible, which is different from fine-tuning that too much data will not gain more performance. Since changing more tokens leads to more semantic shifts, we use the proportion of changed tokens to reflect semantic shift degrees. Then we find that KD prefers augmented data with a larger semantic shift degree (e.g., changing 30% tokens is generally the best option for KD) than fine-tuning (changing 10% tokens). Besides, our findings show that smaller datasets prefer larger degrees until the out-of-distribution problem occurs (e.g., datasets with less than 10k inputs may prefer the 50% degree, and datasets with more than 100k inputs may prefer the 10% degree). Our work sheds light on the preference difference in data augmentation between fine-tuning and knowledge distillation and encourages the community to explore KD-specific data augmentation methods., Comment: 14 pages, 8 tables, 5 figures

Large language Models (LLMs) have achieved promising performance on arithmetic reasoning tasks by incorporating step-by-step chain-of-thought (CoT) prompting. However, LLMs face challenges in maintaining factual consistency during reasoning, exhibiting tendencies to condition overlooking, question misinterpretation, and condition hallucination over given problems. Existing methods use coarse-grained feedback (e.g., whether the answer is correct) to improve factual consistency. In this work, we propose RCoT (Reversing Chain-of-Thought), a novel method to improve LLMs' reasoning abilities by automatically detecting and rectifying factual inconsistency in LLMs, generated solutions. To detect factual inconsistency, RCoT first asks LLMs to reconstruct the problem based on generated solutions. Then fine-grained comparisons between the original problem and the reconstructed problem expose the factual inconsistency in the original solutions. To rectify the solution, RCoT formulates detected factual inconsistency into fine-grained feedback to guide LLMs in revising solutions. Experimental results demonstrate improvements of RCoT over standard CoT, Self-Consistency and Self-Refine across seven arithmetic datasets. Moreover, we find that manually written fine-grained feedback can dramatically improve LLMs' reasoning abilities (e.g., ChatGPT reaches 94.6% accuracy on GSM8K), encouraging the community to further explore the fine-grained feedback generation methods., Comment: 24 pages, 21 figures

Games have been the perfect test-beds for artificial intelligence research for the characteristics that widely exist in real-world scenarios. Learning and optimisation, decision making in dynamic and uncertain environments, game theory, planning and scheduling, design and education are common research areas shared between games and real-world problems. Numerous open-source games or game-based environments have been implemented for studying artificial intelligence. In addition to single- or multi-player, collaborative or adversarial games, there has also been growing interest in implementing platforms for creative design in recent years. Those platforms provide ideal benchmarks for exploring and comparing artificial intelligence ideas and techniques. This paper reviews the games and game-based platforms for artificial intelligence research, provides guidance on matching particular types of artificial intelligence with suitable games for testing and matching particular needs in games with suitable artificial intelligence techniques, discusses the research trend induced by the evolution of those games and platforms, and gives an outlook., Comment: This paper has been accepted by IEEE Transactions on Artificial Intelligence

To address the challenges of reliability analysis in high-dimensional probability spaces, this paper proposes a new metamodeling method that couples active subspace, heteroscedastic Gaussian process, and active learning. The active subspace is leveraged to identify low-dimensional salient features of a high-dimensional computational model. A surrogate computational model is built in the low-dimensional feature space by a heteroscedastic Gaussian process. Active learning adaptively guides the surrogate model training toward the critical region that significantly contributes to the failure probability. A critical trait of the proposed method is that the three main ingredients-active subspace, heteroscedastic Gaussian process, and active learning-are coupled to adaptively optimize the feature space mapping in conjunction with the surrogate modeling. This coupling empowers the proposed method to accurately solve nontrivial high-dimensional reliability problems via low-dimensional surrogate modeling. Finally, numerical examples of a high-dimensional nonlinear function and structural engineering applications are investigated to verify the performance of the proposed method.

Timely and comprehensive understanding of emerging events is crucial for effective decision-making; automating situation report generation can significantly reduce the time, effort, and cost for intelligence analysts. In this work, we identify intelligence analysts' practices and preferences for AI assistance in situation report generation to guide the design strategies for an effective, trust-building interface that aligns with their thought processes and needs. Next, we introduce SmartBook, an automated framework designed to generate situation reports from large volumes of news data, creating structured reports by automatically discovering event-related strategic questions. These reports include multiple hypotheses (claims), summarized and grounded to sources with factual evidence, to promote in-depth situation understanding. Our comprehensive evaluation of SmartBook, encompassing a user study alongside a content review with an editing study, reveals SmartBook's effectiveness in generating accurate and relevant situation reports. Qualitative evaluations indicate over 80% of questions probe for strategic information, and over 90% of summaries produce tactically useful content, being consistently favored over summaries from a large language model integrated with web search. The editing study reveals that minimal information is removed from the generated text (under 2.5%), suggesting that SmartBook provides analysts with a valuable foundation for situation reports, Comment: Preprint