Search

A quasi-local mass, typically defined as an integral over a spacelike $2$-surface $\Sigma$, should encode information about the gravitational field within a finite, extended region bounded by $\Sigma$. Therefore, in attempts to quantize gravity, one may consider an infinite dimensional space of $2$-surfaces instead of an infinite dimensional space of $4$-dimensional Lorentzian spacetimes. However, existing definitions for quasilocal mass only applies to surfaces outside an horizon whose mean curvature vector is spacelike. In this paper, we propose an extension of the Wang-Yau quasi-local energy/mass to surfaces with timelike mean curvature vector, including in particular trapped surfaces. We adopt the same canonical gauge as in the Wang-Yau quasi-local energy but allow the pulled back "killing vector" to the physical spacetime to be spacelike. We define the new quasi-local energy along the Hamiltonian formulation of the Wang-Yau quasi-local energy. The new definition yields a positive definite surface energy density and a new divergence free current. Calculations for coordinate spheres in Kerr family spacetime are shown. In the spherical symmetric case, our definition reduces to a previous definition \cite{lundgren2007self}.

We look at the strong field behavior of the Wang-Yau quasi-local energy. In particular, we examine the limit of the Wang-Yau quasi-local energy as the defining spacelike 2-surface {\Sigma} approaches an apparent horizon from outside. Assuming any smooth isometric embedding, we find that the Wang-Yau quasi-local energy blows up in this limit, unless the horizon can be isometrically embedded into R3. Moreover, if the horizon cannot be embedded into R3, the optimal embedding equation is not solvable for {\Sigma} near the horizon. We discuss the implications of our results in the conclusion section.

We resolve the nature of the quantum phase transition between a N\'eel antiferromagnet and a valence-bond solid in two-dimensional spin-1/2 magnets. We study a class of $J$-$Q$ models, in which Heisenberg exchange $J$ competes with interactions $Q_n$ formed by products of $n$ singlet projectors on adjacent parallel lattice links. QMC simulations provide unambiguous evidence for first-order transitions, with the discontinuities increasing with $n$. For $n=2$ and $n=3$ models, the first-order signatures are very weak. On intermediate length scales, we extract well-defined scaling dimensions (critical exponents) that are common to the models with small $n$, indicating proximity to a quantum critical point. By combining two $Q$ terms, the transition can be tuned from weak to more strongly first-order. The two coexisting orders on the first-order line scale with a large exponent $\beta \approx 0.85$. This exponent and others are close to bounds for an SO($5$) symmetric CFT with a relevant SO($5$) singlet. We characterize the emergent SO($5$) symmetry by the scaling dimensions of its leading irrelevant perturbations. The large $\beta$ value and a large correlation length exponent, $\nu \approx 1.4$, partially explain why the transition remains near-critical even quite far away from the critical point and in many different models without fine-tuning. In addition, we find that few-spin lattice operators are dominated by the SO($5$) violating field (the traceless symmetric tensor), and interactions involving many spins are required to observe strong effects of the relevant SO($5$) singlet. The exponent that had previously been identified with the divergent correlation length when crossing between the two phases does not have a corresponding CFT operator. We explain this emergent pseudocritical scale by a mechanism relying on a dangerously irrelevant SO($5$) perturbation., Comment: 57 pages, 36 figures

We review Wang-Yau quasi-local definitions along the line of gravitational Hamiltonian. This makes clear the connection and difference between Wang-Yau definition and Brown-York or even global ADM definition. We make a brief comment on admissibility condition in Wang-Yau quasi-lcoal mass. We extend the positivity proof for Wang-Yau quasi-local energy to allow possible presence of strictly stable apparent horizons through establishing solvability of Dirac equation in certain 3-manifolds that possess cylindrical ends, as in the case of Jang's graph blowing up at marginally outer trapped surfaces.

Language models (LMs) are trained on web text originating from many points in time and, in general, without any explicit temporal grounding. This work investigates the temporal chaos of pretrained LMs and explores various methods to align their internal knowledge to a target time, which we call "temporal alignment." To do this, we first automatically construct a dataset containing 20K time-sensitive questions and their answers for each year from 2000 to 2023. Based on this dataset, we empirically show that pretrained LMs (e.g., LLaMa2), despite having a recent pretraining cutoff (e.g., 2022), mostly answer questions using earlier knowledge (e.g., in 2019). We then develop several methods, from prompting to finetuning, to align LMs to use their most recent knowledge when answering questions, and investigate various factors in this alignment. Our experiments demonstrate that aligning LLaMa2 to the year 2022 can enhance its performance by up to 62% according to that year's answers. This improvement occurs even without explicitly mentioning time information, indicating the possibility of aligning models' internal sense of time after pretraining. Finally, we find that alignment to a historical time is also possible, with up to 2.8$\times$ the performance of the unaligned LM in 2010 if finetuning models to that year. These findings hint at the sophistication of LMs' internal knowledge organization and the necessity of tuning them properly., Comment: Accepted as Findings of ACL 2024. Our code and data is available at https://github.com/yizhongw/llm-temporal-alignment

Fine-tuning and inference with large Language Models (LM) are generally known to be expensive. Parameter-efficient fine-tuning over pretrained LMs reduces training memory by updating a small number of LM parameters but does not improve inference efficiency. Structured pruning improves LM inference efficiency by removing consistent parameter blocks, yet often increases training memory and time. To improve both training and inference efficiency, we introduce APT that adaptively prunes and tunes parameters for the LMs. At the early stage of fine-tuning, APT dynamically adds salient tuning parameters for fast and accurate convergence while discarding unimportant parameters for efficiency. Compared to baselines, our experiments show that APT maintains up to 98% task performance when pruning RoBERTa and T5 models with 40% parameters left while keeping 86.4% LLaMA models' performance with 70% parameters remained. Furthermore, APT speeds up LMs fine-tuning by up to 8x and reduces large LMs memory training footprint by up to 70%., Comment: Accepted to ICML 2024 Oral; code available at https://github.com/ROIM1998/APT

With the Generative Pre-trained Transformer 3.5 (GPT-3.5) exhibiting remarkable reasoning and comprehension abilities in Natural Language Processing (NLP), most Question Answering (QA) research has primarily centered around general QA tasks based on GPT, neglecting the specific challenges posed by Complex Table QA. In this paper, we propose to incorporate GPT-3.5 to address such challenges, in which complex tables are reconstructed into tuples and specific prompt designs are employed for dialogues. Specifically, we encode each cell's hierarchical structure, position information, and content as a tuple. By enhancing the prompt template with an explanatory description of the meaning of each tuple and the logical reasoning process of the task, we effectively improve the hierarchical structure awareness capability of GPT-3.5 to better parse the complex tables. Extensive experiments and results on Complex Table QA datasets, i.e., the open-domain dataset HiTAB and the aviation domain dataset AIT-QA show that our approach significantly outperforms previous work on both datasets, leading to state-of-the-art (SOTA) performance., Comment: EMNLP 2023 Main

Neural Radiance Fields (NeRF) have significantly advanced the generation of highly realistic and expressive 3D scenes. However, the task of editing NeRF, particularly in terms of geometry modification, poses a significant challenge. This issue has obstructed NeRF's wider adoption across various applications. To tackle the problem of efficiently editing neural implicit fields, we introduce Neural Impostor, a hybrid representation incorporating an explicit tetrahedral mesh alongside a multigrid implicit field designated for each tetrahedron within the explicit mesh. Our framework bridges the explicit shape manipulation and the geometric editing of implicit fields by utilizing multigrid barycentric coordinate encoding, thus offering a pragmatic solution to deform, composite, and generate neural implicit fields while maintaining a complex volumetric appearance. Furthermore, we propose a comprehensive pipeline for editing neural implicit fields based on a set of explicit geometric editing operations. We show the robustness and adaptability of our system through diverse examples and experiments, including the editing of both synthetic objects and real captured data. Finally, we demonstrate the authoring process of a hybrid synthetic-captured object utilizing a variety of editing operations, underlining the transformative potential of Neural Impostor in the field of 3D content creation and manipulation., Comment: Accepted at Pacific Graphics 2023 and Computer Graphics Forum

Secure outsourced computation is critical for cloud computing to safeguard data confidentiality and ensure data usability. Recently, secure outsourced computation schemes following a twin-server architecture based on partially homomorphic cryptosystems have received increasing attention. The Secure Outsourced Computation on Integers (SOCI) [1] toolkit is the state-of-the-art among these schemes which can perform secure computation on integers without requiring the costly bootstrapping operation as in fully homomorphic encryption; however, SOCI suffers from relatively large computation and communication overhead. In this paper, we propose SOCI+ which significantly improves the performance of SOCI. Specifically, SOCI+ employs a novel (2, 2)-threshold Paillier cryptosystem with fast encryption and decryption as its cryptographic primitive, and supports a suite of efficient secure arithmetic computation on integers protocols, including a secure multiplication protocol (SMUL), a secure comparison protocol (SCMP), a secure sign bit-acquisition protocol (SSBA), and a secure division protocol (SDIV), all based on the (2, 2)-threshold Paillier cryptosystem with fast encryption and decryption. In addition, SOCI+ incorporates an offline and online computation mechanism to further optimize its performance. We perform rigorous theoretical analysis to prove the correctness and security of SOCI+. Compared with SOCI, our experimental evaluation shows that SOCI+ is up to 5.4 times more efficient in computation and 40% less in communication overhead.

Identifying a general quasi-local notion of energy-momentum and angular momentum would be an important advance in general relativity with potentially important consequences for mathematical and astrophysical studies in general relativity. In this paper we study a promising approach to this problem first proposed by Wang and Yau in 2009 based on isometric embeddings of closed surfaces in Minkowski space. We study the properties of the Wang-Yau quasi-local mass in high accuracy numerical simulations of the head-on collisions of two non-spinning black holes within full general relativity. We discuss the behavior of the Wang-Yau quasi-local mass on constant expansion surfaces and we compare its behavior with the irreducible mass. We investigate the time evolution of the Wang-Yau Quasi-local mass in numerical examples. In addition we discuss mathematical subtleties in defining the Wang-Yau mass for marginally trapped surfaces.

Crowdsourcing is an emerging computing paradigm that takes advantage of the intelligence of a crowd to solve complex problems effectively. Besides collecting and processing data, it is also a great demand for the crowd to conduct optimization. Inspired by this, this paper intends to introduce crowdsourcing into evolutionary computation (EC) to propose a crowdsourcing-based evolutionary computation (CEC) paradigm for distributed optimization. EC is helpful for optimization tasks of crowdsourcing and in turn, crowdsourcing can break the spatial limitation of EC for large-scale distributed optimization. Therefore, this paper firstly introduces the paradigm of crowdsourcing-based distributed optimization. Then, CEC is elaborated. CEC performs optimization based on a server and a group of workers, in which the server dispatches a large task to workers. Workers search for promising solutions through EC optimizers and cooperate with connected neighbors. To eliminate uncertainties brought by the heterogeneity of worker behaviors and devices, the server adopts the competitive ranking and uncertainty detection strategy to guide the cooperation of workers. To illustrate the satisfactory performance of CEC, a crowdsourcing-based swarm optimizer is implemented as an example for extensive experiments. Comparison results on benchmark functions and a distributed clustering optimization problem demonstrate the potential applications of CEC.

Recently, evolutionary computation (EC) has been promoted by machine learning, distributed computing, and big data technologies, resulting in new research directions of EC like distributed EC and surrogate-assisted EC. These advances have significantly improved the performance and the application scope of EC, but also trigger privacy leakages, such as the leakage of optimal results and surrogate model. Accordingly, evolutionary computation combined with privacy protection is becoming an emerging topic. However, privacy concerns in evolutionary computation lack a systematic exploration, especially for the object, motivation, position, and method of privacy protection. To this end, in this paper, we discuss three typical optimization paradigms (i.e., \textit{centralized optimization, distributed optimization, and data-driven optimization}) to characterize optimization modes of evolutionary computation and propose BOOM to sort out privacy concerns in evolutionary computation. Specifically, the centralized optimization paradigm allows clients to outsource optimization problems to the centralized server and obtain optimization solutions from the server. While the distributed optimization paradigm exploits the storage and computational power of distributed devices to solve optimization problems. Also, the data-driven optimization paradigm utilizes data collected in history to tackle optimization problems lacking explicit objective functions. Particularly, this paper adopts BOOM to characterize the object and motivation of privacy protection in three typical optimization paradigms and discusses potential privacy-preserving technologies balancing optimization performance and privacy guarantees in three typical optimization paradigms. Furthermore, this paper attempts to foresee some new research directions of privacy-preserving evolutionary computation.

Automatic unknown word detection techniques can enable new applications for assisting English as a Second Language (ESL) learners, thus improving their reading experiences. However, most modern unknown word detection methods require dedicated eye-tracking devices with high precision that are not easily accessible to end-users. In this work, we propose GazeReader, an unknown word detection method only using a webcam. GazeReader tracks the learner's gaze and then applies a transformer-based machine learning model that encodes the text information to locate the unknown word. We applied knowledge enhancement including term frequency, part of speech, and named entity recognition to improve the performance. The user study indicates that the accuracy and F1-score of our method were 98.09% and 75.73%, respectively. Lastly, we explored the design scope for ESL reading and discussed the findings., Comment: This paper has been accepted by ACM CHI 2023

As an essential technology underpinning trusted computing, the trusted execution environment (TEE) allows one to launch computation tasks on both on- and off-premises data while assuring confidentiality and integrity. This article provides a systematic review and comparison of TEE-based secure computation protocols. We first propose a taxonomy that classifies secure computation protocols into three major categories, namely secure outsourced computation, secure distributed computation and secure multi-party computation. To enable a fair comparison of these protocols, we also present comprehensive assessment criteria with respect to four aspects: setting, methodology, security and performance. Based on these criteria, we review, discuss and compare the state-of-the-art TEE-based secure computation protocols for both general-purpose computation functions and special-purpose ones, such as privacy-preserving machine learning and encrypted database queries. To the best of our knowledge, this article is the first survey to review TEE-based secure computation protocols and the comprehensive comparison can serve as a guideline for selecting suitable protocols for deployment in practice. Finally, we also discuss several future research directions and challenges.

Dataset biases are notoriously detrimental to model robustness and generalization. The identify-emphasize paradigm appears to be effective in dealing with unknown biases. However, we discover that it is still plagued by two challenges: A, the quality of the identified bias-conflicting samples is far from satisfactory; B, the emphasizing strategies only produce suboptimal performance. In this paper, for challenge A, we propose an effective bias-conflicting scoring method (ECS) to boost the identification accuracy, along with two practical strategies -- peer-picking and epoch-ensemble. For challenge B, we point out that the gradient contribution statistics can be a reliable indicator to inspect whether the optimization is dominated by bias-aligned samples. Then, we propose gradient alignment (GA), which employs gradient statistics to balance the contributions of the mined bias-aligned and bias-conflicting samples dynamically throughout the learning process, forcing models to leverage intrinsic features to make fair decisions. Furthermore, we incorporate self-supervised (SS) pretext tasks into training, which enable models to exploit richer features rather than the simple shortcuts, resulting in more robust models. Experiments are conducted on multiple datasets in various settings, demonstrating that the proposed solution can mitigate the impact of unknown biases and achieve state-of-the-art performance.

Fully test-time adaptation aims at adapting a pre-trained model to the test stream during real-time inference, which is urgently required when the test distribution differs from the training distribution. Several efforts have been devoted to improving adaptation performance. However, we find that two unfavorable defects are concealed in the prevalent adaptation methodologies like test-time batch normalization (BN) and self-learning. First, we reveal that the normalization statistics in test-time BN are completely affected by the currently received test samples, resulting in inaccurate estimates. Second, we show that during test-time adaptation, the parameter update is biased towards some dominant classes. In addition to the extensively studied test stream with independent and class-balanced samples, we further observe that the defects can be exacerbated in more complicated test environments, such as (time) dependent or class-imbalanced data. We observe that previous approaches work well in certain scenarios while show performance degradation in others due to their faults. In this paper, we provide a plug-in solution called DELTA for Degradation-freE fuLly Test-time Adaptation, which consists of two components: (i) Test-time Batch Renormalization (TBR), introduced to improve the estimated normalization statistics. (ii) Dynamic Online re-weighTing (DOT), designed to address the class bias within optimization. We investigate various test-time adaptation methods on three commonly used datasets with four scenarios, and a newly introduced real-world dataset. DELTA can help them deal with all scenarios simultaneously, leading to SOTA performance.

Web and artificial intelligence technologies, especially semantic web and knowledge graph (KG), have recently raised significant attention in educational scenarios. Nevertheless, subject-specific KGs for K-12 education still lack sufficiency and sustainability from knowledge and data perspectives. To tackle these issues, we propose EDUKG, a heterogeneous sustainable K-12 Educational Knowledge Graph. We first design an interdisciplinary and fine-grained ontology for uniformly modeling knowledge and resource in K-12 education, where we define 635 classes, 445 object properties, and 1314 datatype properties in total. Guided by this ontology, we propose a flexible methodology for interactively extracting factual knowledge from textbooks. Furthermore, we establish a general mechanism based on our proposed generalized entity linking system for EDUKG's sustainable maintenance, which can dynamically index numerous heterogeneous resources and data with knowledge topics in EDUKG. We further evaluate EDUKG to illustrate its sufficiency, richness, and variability. We publish EDUKG with more than 252 million entities and 3.86 billion triplets. Our code and data repository is now available at https://github.com/THU-KEG/EDUKG., Comment: 17 pages, 8 figures

Partial label learning (PLL) learns from training examples each associated with multiple candidate labels, among which only one is valid. In recent years, benefiting from the strong capability of dealing with ambiguous supervision and the impetus of modern data augmentation methods, consistency regularization-based PLL methods have achieved a series of successes and become mainstream. However, as the partial annotation becomes insufficient, their performances drop significantly. In this paper, we leverage easily accessible unlabeled examples to facilitate the partial label consistency regularization. In addition to a partial supervised loss, our method performs a controller-guided consistency regularization at both the label-level and representation-level with the help of unlabeled data. To minimize the disadvantages of insufficient capabilities of the initial supervised model, we use the controller to estimate the confidence of each current prediction to guide the subsequent consistency regularization. Furthermore, we dynamically adjust the confidence thresholds so that the number of samples of each class participating in consistency regularization remains roughly equal to alleviate the problem of class-imbalance. Experiments show that our method achieves satisfactory performances in more practical situations, and its modules can be applied to existing PLL methods to enhance their capabilities.

Deep models have been widely and successfully used in image manipulation detection, which aims to classify tampered images and localize tampered regions. Most existing methods mainly focus on extracting global features from tampered images, while neglecting the relationships of local features between tampered and authentic regions within a single tampered image. To exploit such spatial relationships, we propose Proposal Contrastive Learning (PCL) for effective image manipulation detection. Our PCL consists of a two-stream architecture by extracting two types of global features from RGB and noise views respectively. To further improve the discriminative power, we exploit the relationships of local features through a proxy proposal contrastive learning task by attracting/repelling proposal-based positive/negative sample pairs. Moreover, we show that our PCL can be easily adapted to unlabeled data in practice, which can reduce manual labeling costs and promote more generalizable features. Extensive experiments among several standard datasets demonstrate that our PCL can be a general module to obtain consistent improvement. The code is available at https://github.com/Sandy-Zeng/PCL.

Attention-based models have been widely used in many areas, such as computer vision and natural language processing. However, relevant applications in time series classification (TSC) have not been explored deeply yet, causing a significant number of TSC algorithms still suffer from general problems of attention mechanism, like quadratic complexity. In this paper, we promote the efficiency and performance of the attention mechanism by proposing our flexible multi-head linear attention (FMLA), which enhances locality awareness by layer-wise interactions with deformable convolutional blocks and online knowledge distillation. What's more, we propose a simple but effective mask mechanism that helps reduce the noise influence in time series and decrease the redundancy of the proposed FMLA by masking some positions of each given series proportionally. To stabilize this mechanism, samples are forwarded through the model with random mask layers several times and their outputs are aggregated to teach the same model with regular mask layers. We conduct extensive experiments on 85 UCR2018 datasets to compare our algorithm with 11 well-known ones and the results show that our algorithm has comparable performance in terms of top-1 accuracy. We also compare our model with three Transformer-based models with respect to the floating-point operations per second and number of parameters and find that our algorithm achieves significantly better efficiency with lower complexity.

Evolutionary algorithms (EAs), such as the genetic algorithm (GA), offer an elegant way to handle combinatorial optimization problems (COPs). However, limited by expertise and resources, most users do not have enough capability to implement EAs to solve COPs. An intuitive and promising solution is to outsource evolutionary operations to a cloud server, whilst it suffers from privacy concerns. To this end, this paper proposes a novel computing paradigm, evolution as a service (EaaS), where a cloud server renders evolutionary computation services for users without sacrificing users' privacy. Inspired by the idea of EaaS, this paper designs PEGA, a novel privacy-preserving GA for COPs. Specifically, PEGA enables users outsourcing COPs to the cloud server holding a competitive GA and approximating the optimal solution in a privacy-preserving manner. PEGA features the following characteristics. First, any user without expertise and enough resources can solve her COPs. Second, PEGA does not leak contents of optimization problems, i.e., users' privacy. Third, PEGA has the same capability as the conventional GA to approximate the optimal solution. We implements PEGA falling in a twin-server architecture and evaluates it in the traveling salesman problem (TSP, a widely known COP). Particularly, we utilize encryption cryptography to protect users' privacy and carefully design a suit of secure computing protocols to support evolutionary operators of GA on encrypted data. Privacy analysis demonstrates that PEGA does not disclose the contents of the COP to the cloud server. Experimental evaluation results on four TSP datasets show that PEGA is as effective as the conventional GA in approximating the optimal solution.

Multi-access edge computing (MEC) and network function virtualization (NFV) are promising technologies to support emerging IoT applications, especially those computation-intensive. In NFV-enabled MEC environment, service function chain (SFC), i.e., a set of ordered virtual network functions (VNFs), can be mapped on MEC servers. Mobile devices (MDs) can offload computation-intensive applications, which can be represented by SFCs, fully or partially to MEC servers for remote execution. This paper studies the partial offloading and SFC mapping joint optimization (POSMJO) problem in an NFV-enabled MEC system, where an incoming task can be partitioned into two parts, one for local execution and the other for remote execution. The objective is to minimize the average cost in the long term which is a combination of execution delay, MD's energy consumption, and usage charge for edge computing. This problem consists of two closely related decision-making steps, namely task partition and VNF placement, which is highly complex and quite challenging. To address this, we propose a cooperative dual-agent deep reinforcement learning (CDADRL) algorithm, where we design a framework enabling interaction between two agents. Simulation results show that the proposed algorithm outperforms three combinations of deep reinforcement learning algorithms in terms of cumulative and average episodic rewards and it overweighs a number of baseline algorithms with respect to execution delay, energy consumption, and usage charge.

This paper studies the trajectory control and task offloading (TCTO) problem in an unmanned aerial vehicle (UAV)-assisted mobile edge computing system, where a UAV flies along a planned trajectory to collect computation tasks from smart devices (SDs). We consider a scenario that SDs are not directly connected by the base station (BS) and the UAV has two roles to play: MEC server or wireless relay. The UAV makes task offloading decisions online, in which the collected tasks can be executed locally on the UAV or offloaded to the BS for remote processing. The TCTO problem involves multi-objective optimization as its objectives are to minimize the task delay and the UAV's energy consumption, and maximize the number of tasks collected by the UAV, simultaneously. This problem is challenging because the three objectives conflict with each other. The existing reinforcement learning (RL) algorithms, either single-objective RLs or single-policy multi-objective RLs, cannot well address the problem since they cannot output multiple policies for various preferences (i.e. weights) across objectives in a single run. This paper adapts the evolutionary multi-objective RL (EMORL), a multi-policy multi-objective RL, to the TCTO problem. This algorithm can output multiple optimal policies in just one run, each optimizing a certain preference. The simulation results demonstrate that the proposed algorithm can obtain more excellent nondominated policies by striking a balance between the three objectives regarding policy quality, compared with two evolutionary and two multi-policy RL algorithms.

This paper proposes an efficient federated distillation learning system (EFDLS) for multi-task time series classification (TSC). EFDLS consists of a central server and multiple mobile users, where different users may run different TSC tasks. EFDLS has two novel components, namely a feature-based student-teacher (FBST) framework and a distance-based weights matching (DBWM) scheme. Within each user, the FBST framework transfers knowledge from its teacher's hidden layers to its student's hidden layers via knowledge distillation, with the teacher and student having identical network structure. For each connected user, its student model's hidden layers' weights are uploaded to the EFDLS server periodically. The DBWM scheme is deployed on the server, with the least square distance used to measure the similarity between the weights of two given models. This scheme finds a partner for each connected user such that the user's and its partner's weights are the closest among all the weights uploaded. The server exchanges and sends back the user's and its partner's weights to these two users which then load the received weights to their teachers' hidden layers. Experimental results show that the proposed EFDLS achieves excellent performance on a set of selected UCR2018 datasets regarding top-1 accuracy., Comment: 11 pages

Models notoriously suffer from dataset biases which are detrimental to robustness and generalization. The identify-emphasize paradigm shows a promising effect in dealing with unknown biases. However, we find that it is still plagued by two challenges: A, the quality of the identified bias-conflicting samples is far from satisfactory; B, the emphasizing strategies just yield suboptimal performance. In this work, for challenge A, we propose an effective bias-conflicting scoring method to boost the identification accuracy with two practical strategies -- peer-picking and epoch-ensemble. For challenge B, we point out that the gradient contribution statistics can be a reliable indicator to inspect whether the optimization is dominated by bias-aligned samples. Then, we propose gradient alignment, which employs gradient statistics to balance the contributions of the mined bias-aligned and bias-conflicting samples dynamically throughout the learning process, forcing models to leverage intrinsic features to make fair decisions. Experiments are conducted on multiple datasets in various settings, demonstrating that the proposed solution can alleviate the impact of unknown biases and achieve state-of-the-art performance.