arXiv subjects

Jialu Wang

Publications and source records attributed to Jialu Wang.

At least 55 records · Page 3Linked to original sources

arXiv2024

Unmasking and Improving Data Credibility: A Study with Datasets for Training Harmless Language Models

Language models have shown promise in various tasks but can be affected by undesired data during training, fine-tuning, or alignment. For example, if some unsafe conversations are wrongly annotated as safe ones, the model fine-tuned on these samples may be harmful. Therefore, the correctness of annotations, i.e., the credibility of the dataset, is important. This study focuses on the credibility of real-world datasets, including the popular benchmarks Jigsaw Civil Comments, Anthropic Harmless & Red Team, PKU BeaverTails & SafeRLHF, that can be used for training a harmless language model. Given the cost and difficulty of cleaning these datasets by humans, we introduce a systematic framework for evaluating the credibility of datasets, identifying label errors, and evaluating the influence of noisy labels in the curated language data, specifically focusing on unsafe comments and conversation classification. With the framework, we find and fix an average of 6.16% label errors in 11 datasets constructed from the above benchmarks. The data credibility and downstream learning performance can be remarkably improved by directly fixing label errors, indicating the significance of cleaning existing real-world datasets. We provide an open-source tool, Docta, for data cleaning at https://github.com/Docta-ai/docta.

cs.LG↗

arXiv2024

WSCLoc: Weakly-Supervised Sparse-View Camera Relocalization

Despite the advancements in deep learning for camera relocalization tasks, obtaining ground truth pose labels required for the training process remains a costly endeavor. While current weakly supervised methods excel in lightweight label generation, their performance notably declines in scenarios with sparse views. In response to this challenge, we introduce WSCLoc, a system capable of being customized to various deep learning-based relocalization models to enhance their performance under weakly-supervised and sparse view conditions. This is realized with two stages. In the initial stage, WSCLoc employs a multilayer perceptron-based structure called WFT-NeRF to co-optimize image reconstruction quality and initial pose information. To ensure a stable learning process, we incorporate temporal information as input. Furthermore, instead of optimizing SE(3), we opt for $\mathfrak{sim}(3)$ optimization to explicitly enforce a scale constraint. In the second stage, we co-optimize the pre-trained WFT-NeRF and WFT-Pose. This optimization is enhanced by Time-Encoding based Random View Synthesis and supervised by inter-frame geometric constraints that consider pose, depth, and RGB information. We validate our approaches on two publicly available datasets, one outdoor and one indoor. Our experimental results demonstrate that our weakly-supervised relocalization solutions achieve superior pose estimation accuracy in sparse-view scenarios, comparable to state-of-the-art camera relocalization methods. We will make our code publicly available.

cs.CV↗

arXiv2023

Parameter-Efficient Cross-lingual Transfer of Vision and Language Models via Translation-based Alignment

Pre-trained vision and language models such as CLIP have witnessed remarkable success in connecting images and texts with a primary focus on English texts. Despite recent efforts to extend CLIP to support other languages, disparities in performance among different languages have been observed due to uneven resource availability. Additionally, current cross-lingual transfer methods of those pre-trained models would consume excessive resources for a large number of languages. Therefore, we propose a new parameter-efficient cross-lingual transfer learning framework that utilizes a translation-based alignment method to mitigate multilingual disparities and explores parameter-efficient fine-tuning methods for parameter-efficient cross-lingual transfer. Extensive experiments on XTD and Multi30K datasets, covering 11 languages under zero-shot, few-shot, and full-dataset learning scenarios, show that our framework significantly reduces the multilingual disparities among languages and improves cross-lingual transfer results, especially in low-resource scenarios, while only keeping and fine-tuning an extremely small number of parameters compared to the full model (e.g., Our framework only requires 0.16\% additional parameters of a full-model for each language in the few-shot learning scenario). The codes are available at \url{https://github.com/eric-ai-lab/PECTVLM}. The codes are available at \url{https://github.com/eric-ai-lab/PECTVLM}.

cs.CL↗

arXiv2023

Clues to potential dipolar-Kondo and RKKY interactions in a polar metal

The coexistence of electric dipoles and itinerant electrons in a solid was postulated decades ago, before being experimentally established in several 'polar metals' during the last decade. Here, we report a concentration-driven polar-to-nonpolar phase transition in electron-doped BaTiO_3. Comparing our case with other polar metals, we find a particular threshold concentration (n*) linked to the dipole density (n_d). The universal ratio n_d/n*=8(0.6) suggests a common mechanism across different polar systems, possibly explained by a dipolar Ruderman-Kittel-Kasuya-Yosida theory. Moreover, in BaTiO_3, we observe enhanced thermopower and upturn on resistivity at low temperatures near n*, resembling the Kondo effect. We argue that local electric dipoles act as two-level-systems, whose fluctuations couple with surrounding electron clouds, giving rise to a potential dipolar-counterpart of the Kondo effect. Our findings unveil a mostly uncharted territory for exploring emerging physics associated with electron-dipole correlations, encouraging further theoretical work on dipolar-RKKY and Kondo interactions.

cond-mat.str-el↗

arXiv2023

T2IAT: Measuring Valence and Stereotypical Biases in Text-to-Image Generation

Warning: This paper contains several contents that may be toxic, harmful, or offensive. In the last few years, text-to-image generative models have gained remarkable success in generating images with unprecedented quality accompanied by a breakthrough of inference speed. Despite their rapid progress, human biases that manifest in the training examples, particularly with regard to common stereotypical biases, like gender and skin tone, still have been found in these generative models. In this work, we seek to measure more complex human biases exist in the task of text-to-image generations. Inspired by the well-known Implicit Association Test (IAT) from social psychology, we propose a novel Text-to-Image Association Test (T2IAT) framework that quantifies the implicit stereotypes between concepts and valence, and those in the images. We replicate the previously documented bias tests on generative models, including morally neutral tests on flowers and insects as well as demographic stereotypical tests on diverse social attributes. The results of these experiments demonstrate the presence of complex stereotypical behaviors in image generations.

cs.CL↗

arXiv2023

Deep Learning based Multi-Label Image Classification of Protest Activities

With the rise of internet technology amidst increasing rates of urbanization, sharing information has never been easier thanks to globally-adopted platforms for digital communication. The resulting output of massive amounts of user-generated data can be used to enhance our understanding of significant societal issues particularly for urbanizing areas. In order to better analyze protest behavior, we enhanced the GSR dataset and manually labeled all the images. We used deep learning techniques to analyze social media data to detect social unrest through image classification, which performed good in predict multi-attributes, then also used map visualization to display protest behaviors across the country.

cs.CV↗

arXiv2022

Fairness Transferability Subject to Bounded Distribution Shift

Given an algorithmic predictor that is "fair" on some source distribution, will it still be fair on an unknown target distribution that differs from the source within some bound? In this paper, we study the transferability of statistical group fairness for machine learning predictors (i.e., classifiers or regressors) subject to bounded distribution shifts. Such shifts may be introduced by initial training data uncertainties, user adaptation to a deployed predictor, dynamic environments, or the use of pre-trained models in new settings. Herein, we develop a bound that characterizes such transferability, flagging potentially inappropriate deployments of machine learning for socially consequential tasks. We first develop a framework for bounding violations of statistical fairness subject to distribution shift, formulating a generic upper bound for transferred fairness violations as our primary result. We then develop bounds for specific worked examples, focusing on two commonly used fairness definitions (i.e., demographic parity and equalized odds) and two classes of distribution shift (i.e., covariate shift and label shift). Finally, we compare our theoretical bounds to deterministic models of distribution shift and against real-world data, finding that we are able to estimate fairness violation bounds in practice, even when simplifying assumptions are only approximately satisfied.

cs.LG↗

arXiv2022

Utilizing Win Ratio Approaches and Two-Stage Enrichment Designs for Small-Sized Clinical Trials

Conventional methods for analyzing composite endpoints in clinical trials often only focus on the time to the first occurrence of all events in the composite. Therefore, they have inherent limitations because the individual patients' first event can be the outcome of lesser clinical importance. To overcome this limitation, the concept of the win ratio (WR), which accounts for the relative priorities of the components and gives appropriate priority to the more clinically important event, was examined. For example, because mortality has a higher priority than hospitalization, it is reasonable to give a higher priority when obtaining the WR. In this paper, we evaluate three innovative WR methods (stratified matched, stratified unmatched, and unstratified unmatched) for two and multiple components under binary and survival composite endpoints. We compare these methods to traditional ones, including the Cox regression, O'Brien's rank-sum-type test, and the contingency table for controlling study Type I error rate. We also incorporate these approaches into two-stage enrichment designs with the possibility of sample size adaptations to gain efficiency for rare disease studies.

stat.ME↗

arXiv2022

Understanding Instance-Level Impact of Fairness Constraints

A variety of fairness constraints have been proposed in the literature to mitigate group-level statistical bias. Their impacts have been largely evaluated for different groups of populations corresponding to a set of sensitive attributes, such as race or gender. Nonetheless, the community has not observed sufficient explorations for how imposing fairness constraints fare at an instance level. Building on the concept of influence function, a measure that characterizes the impact of a training example on the target model and its predictive performance, this work studies the influence of training examples when fairness constraints are imposed. We find out that under certain assumptions, the influence function with respect to fairness constraints can be decomposed into a kernelized combination of training examples. One promising application of the proposed fairness influence function is to identify suspicious training examples that may cause model discrimination by ranking their influence scores. We demonstrate with extensive experiments that training on a subset of weighty data examples leads to lower fairness violations with a trade-off of accuracy.

cs.LG↗

arXiv2022

Beyond Images: Label Noise Transition Matrix Estimation for Tasks with Lower-Quality Features

The label noise transition matrix, denoting the transition probabilities from clean labels to noisy labels, is crucial for designing statistically robust solutions. Existing estimators for noise transition matrices, e.g., using either anchor points or clusterability, focus on computer vision tasks that are relatively easier to obtain high-quality representations. We observe that tasks with lower-quality features fail to meet the anchor-point or clusterability condition, due to the coexistence of both uninformative and informative representations. To handle this issue, we propose a generic and practical information-theoretic approach to down-weight the less informative parts of the lower-quality features. This improvement is crucial to identifying and estimating the label noise transition matrix. The salient technical challenge is to compute the relevant information-theoretical metrics using only noisy labels instead of clean ones. We prove that the celebrated $f$-mutual information measure can often preserve the order when calculated using noisy labels. We then build our transition matrix estimator using this distilled version of features. The necessity and effectiveness of the proposed method are also demonstrated by evaluating the estimation error on a varied set of tabular data and text classification tasks with lower-quality features. Code is available at github.com/UCSC-REAL/BeyondImages.

cs.LG↗

arXiv2022

Assessing Multilingual Fairness in Pre-trained Multimodal Representations

Recently pre-trained multimodal models, such as CLIP, have shown exceptional capabilities towards connecting images and natural language. The textual representations in English can be desirably transferred to multilingualism and support downstream multimodal tasks for different languages. Nevertheless, the principle of multilingual fairness is rarely scrutinized: do multilingual multimodal models treat languages equally? Are their performances biased towards particular languages? To answer these questions, we view language as the fairness recipient and introduce two new fairness notions, multilingual individual fairness and multilingual group fairness, for pre-trained multimodal models. Multilingual individual fairness requires that text snippets expressing similar semantics in different languages connect similarly to images, while multilingual group fairness requires equalized predictive performance across languages. We characterize the extent to which pre-trained multilingual vision-and-language representations are individually fair across languages. However, extensive experiments demonstrate that multilingual representations do not satisfy group fairness: (1) there is a severe multilingual accuracy disparity issue; (2) the errors exhibit biases across languages conditioning the group of people in the images, including race, gender and age.

cs.CL↗

arXiv2021

Commensurate stacking phase transitions in an intercalated transition metal dichalcogenide

Intercalation and stacking order modulation are two active ways in manipulating the interlayer interaction of transition metal dichalcogenides (TMDCs), which lead to a variety of emergent phases and allow for engineering material properties. Herein, the growth of Pb intercalated TMDCs--Pb(Ta$_{1+x}$Se$_2$)$_2$, the first 124-phase, is reported. Pb(Ta$_{1+x}$Se$_2$)$_2$ exhibits a unique two-step first-order structural phase transition at around 230 K. The transitions are solely associated with the stacking degree of freedom, evolving from a high temperature phase with ABC stacking and symmetry $R3m$ to an intermediate phase with AB stacking and $P3m1$, and finally to a low temperature phase with again symmetry $R3m$, but with ACB stacking. Each step involves a rigid slide of building blocks by a vector [1/3, 2/3, 0]. Intriguingly, gigantic lattice contractions occur at the transitions on warming. At low temperature, bulk superconductivity with $T_\textrm{c}\approx$ 1.8 K is observed. The underlying physics of the structural phase transitions are discussed from first-principle calculations. The symmetry analysis reveals topological nodal lines in the band structure. Our results demonstrate the possibility to realize higher order metal intercalated phases of TMDCs, advance our knowledge of polymorphic transitions and may inspire stacking order engineering in TMDCs and beyond.

cond-mat.mtrl-sci↗

arXiv2021

RADA: Robust Adversarial Data Augmentation for Camera Localization in Challenging Weather

Camera localization is a fundamental and crucial problem for many robotic applications. In recent years, using deep-learning for camera-based localization has become a popular research direction. However, they lack robustness to large domain shifts, which can be caused by seasonal or illumination changes between training and testing data sets. Data augmentation is an attractive approach to tackle this problem, as it does not require additional data to be provided. However, existing augmentation methods blindly perturb all pixels and therefore cannot achieve satisfactory performance. To overcome this issue, we proposed RADA, a system whose aim is to concentrate on perturbing the geometrically informative parts of the image. As a result, it learns to generate minimal image perturbations that are still capable of perplexing the network. We show that when these examples are utilized as augmentation, it greatly improves robustness. We show that our method outperforms previous augmentation techniques and achieves up to two times higher accuracy than the SOTA localization models (e.g., AtLoc and MapNet) when tested on `unseen' challenging weather conditions.

cs.CV↗

arXiv2021

Can Less be More? When Increasing-to-Balancing Label Noise Rates Considered Beneficial

In this paper, we answer the question of when inserting label noise (less informative labels) can instead return us more accurate and fair models. We are primarily inspired by three observations: 1) In contrast to reducing label noise rates, increasing the noise rates is easy to implement; 2) Increasing a certain class of instances' label noise to balance the noise rates (increasing-to-balancing) results in an easier learning problem; 3) Increasing-to-balancing improves fairness guarantees against label bias. In this paper, we first quantify the trade-offs introduced by increasing a certain group of instances' label noise rate w.r.t. the loss of label informativeness and the lowered learning difficulties. We analytically demonstrate when such an increase is beneficial, in terms of either improved generalization power or the fairness guarantees. Then we present a method to insert label noise properly for the task of learning with noisy labels, either without or with a fairness constraint. The primary technical challenge we face is due to the fact that we would not know which data instances are suffering from higher noise, and we would not have the ground truth labels to verify any possible hypothesis. We propose a detection method that informs us which group of labels might suffer from higher noise without using ground truth labels. We formally establish the effectiveness of the proposed solution and demonstrate it with extensive experiments.

cs.LG↗

arXiv2021

Are Gender-Neutral Queries Really Gender-Neutral? Mitigating Gender Bias in Image Search

Internet search affects people's cognition of the world, so mitigating biases in search results and learning fair models is imperative for social good. We study a unique gender bias in image search in this work: the search images are often gender-imbalanced for gender-neutral natural language queries. We diagnose two typical image search models, the specialized model trained on in-domain datasets and the generalized representation model pre-trained on massive image and text data across the internet. Both models suffer from severe gender bias. Therefore, we introduce two novel debiasing approaches: an in-processing fair sampling method to address the gender imbalance issue for training models, and a post-processing feature clipping method base on mutual information to debias multimodal representations of pre-trained models. Extensive experiments on MS-COCO and Flickr30K benchmarks show that our methods significantly reduce the gender bias in image search models.

cs.CV↗

arXiv2021

Thermoelectric properties of Bi2O2Se single crystals

Bismuth oxyselenide (Bi$_2$O$_2$Se) attracts great interest as a potential n-type complement to p-type thermoelectric oxides in practical applications. Previous investigations were generally focused on polycrystals. Here, we performed a study on the thermoelectric properties of Bi$_2$O$_2$Se single crystals. Our samples exhibit electron mobility as high as 250 cm$^2.$V$^{-1}$.s$^{-1}$ and thermal conductivity as low as $2$ W.m$^{-1}$.K$^{-1}$ near room temperature. The maximized figure of merit is yielded to be 0.188 at 390 K, higher than that of polycrystals. Consequently, a rough estimation of the phonon mean free path ($\ell_\textrm{ph}$) from the kinetic model amounts to 12 $Å$ at 390 K and follows a $T^{-1}$ behavior. An extrapolation of $\ell_\textrm{ph}$ to higher temperatures indicates that this system approaches the Ioffe-Regel limit at about 1100 K. In light of the phonon dispersions, we argue that the ultralow $\ell_\textrm{ph}$ is attributed to intense anharmonic phonon-phonon scattering, including Umklapp process and acoustic to optical phonon scattering. Our results suggest that single crystals provide a further improvement of thermoelectric performance of Bi$_2$O$_2$Se.

cond-mat.mtrl-sci↗

arXiv2021

Linear Classifiers that Encourage Constructive Adaptation

Machine learning systems are often used in settings where individuals adapt their features to obtain a desired outcome. In such settings, strategic behavior leads to a sharp loss in model performance in deployment. In this work, we aim to address this problem by learning classifiers that encourage decision subjects to change their features in a way that leads to improvement in both predicted \emph{and} true outcome. We frame the dynamics of prediction and adaptation as a two-stage game, and characterize optimal strategies for the model designer and its decision subjects. In benchmarks on simulated and real-world datasets, we find that classifiers trained using our method maintain the accuracy of existing approaches while inducing higher levels of improvement and less manipulation.

cs.LG↗

arXiv2021

Anisotropic superconductivity in topological crystalline metal Pb$_{1/3}$TaS$_2$ with multiple Dirac fermions

Topological crystalline metals/semimetals (TCMs) have stimulated a great research interest, which broaden the classification of topological phases and provide a valuable platform to explore topological superconductivity. Here, we report the discovery of superconductivity and topological features in Pb-intercalated transition-metal dichalcogenide Pb$_{1/3}$TaS$_2$. Systematic measurements indicate that Pb$_{1/3}$TaS$_2$ is a quasi-two-dimensional (q-2D) type-II superconductor ({\em T}$_c \approx$ 2.8 K) with a significantly enhanced anisotropy of upper critical field ($γ_{H_{c2}}$ = $H_{c2}^{ab}/H_{c2}^{c}$ $\approx$ 17). In addition, first-principles calculations reveal that Pb$_{1/3}$TaS$_2$ hosts multiple topological Dirac fermions in the electronic band structure. We discover four groups of Dirac nodal lines on the $k_z = π$ plane and two sets of Dirac points on the rotation/screw axes, which are protected by crystalline symmetries and robust against spin-orbit coupling (SOC). Dirac-cone-like surface states emerge on the (001) surface because of band inversion. Our work shows that the TCM candidate Pb$_{1/3}$TaS$_2$ is a promising arena to study the interplay between superconductivity and topological Dirac fermions.

cond-mat.supr-con↗