Search

Due to the advancements of modern technologies, large-scale and high-dimensional data have been widely collected in almost every scientific disciplines.This introduces a several challenges including that the data are often accompanied by outliers due to possible measurement error, or many variables follow heavy-tailed distributions. To address these challenges, my thesis proposes methodologies in the setting of the mean estimation and matrix recovery when the data have asymmetric and heavy-tailed distributions. Additionally, I explore the characterization of tail behavior in random outcomes, focusing on expected shortfall, which is widely recognized as a measure of risk. I propose nonparametric approaches for estimating expected shortfall, aiming to enhance its accuracy and applicability.In Chapter 1, we propose a robust estimator to recover approximately low-rank matrices in the presence of heavy-tailed and asymmetric noises.Focusing on three archetypal applications including matrix compressed sensing, matrix completion and multitask learning, we provide sub-Gaussian-type deviation bounds when the noise variables only have bounded variances. Computationally, we propose a matrix version of the local adaptive majorize-minimization algorithm, which is much faster than the alternating direction method of multiplier used in previous work and is scalable to large datasets.Chapter 2 studies the problem of robust and differentially private mean estimation and inference.We first provide a comprehensive analysis of the Huber mean estimator with increasing dimensions, including non-asymptotic deviation bound, Bahadur representation, and (uniform) Gaussian approximations. Then, we privatize the Huber mean estimator via noisy gradient descent, and construct private confidence intervals for the proposed estimator by incorporating a private and robust covariance estimator.In Chapter 3, we consider the problem of nonparametric estimation of conditional expected shortfall functions

Yu-Xian Tang,1,* Zhi-Wei Fan,1,* Jing Li,2,* Hao-Dong Pan,3 Wen-Xin Su,1 Yusanjan Matniyaz,2 Hai-Tao Zhang,4 Yuan-Xi Luo,4 Zhi-Kang Lv,2 Wen-Zhe Wang,2 Ya-Xuan Gao,2 Tuo Pan,2,4 Wan-Zi Xu,1,2 Dong-Jin Wang1– 4 1Department of Cardiac Surgery, Nanjing Drum Tower Hospital, Nanjing Drum Tower Hospital Clinical College, Nanjing University of Chinese Medicine, Nanjing, People’s Republic of China; 2Department of Cardiac Surgery, Nanjing Drum Tower Hospital, Affiliated Hospital of Medical School, Nanjing University, Nanjing, People’s Republic of China; 3Department of Clinical Medicine, Norman Bethune Health Science Center of Jilin University, Changchun, People’s Republic of China; 4Department of Cardiac Surgery, Nanjing Drum Tower Hospital, Peking Union Medical College & Chinese Academy of Medical Sciences, Graduate School of Peking Union Medical College, Nanjing, People’s Republic of China*These authors contributed equally to this workCorrespondence: Dong-Jin Wang; Wan-Zi Xu, Department of Cardiac surgery, Nanjing Drum Tower Hospital, No. 321 Zhongshan Road, Nanjing, Jiangsu, 210008, People’s Republic of China, Email dongjin_wang@126.com; 582321957@qq.comBackground: Sivelestat, a neutrophil elastase inhibitor, is specifically developed to mitigate the occurrence of acute lung injury (ALI) in individuals who are undergoing cardiovascular surgery. However, its impact on patients who are at a heightened risk of developing ALI after scheduled cardiac surgery has yet to be determined. In order to address this knowledge gap, we undertook a study to assess the efficacy of sivelestat in protecting the lungs of these patients.Methods: We conducted a prospective cohort study involving 718 patients who were at high risk of developing postoperative acute lung injury (ALI) and underwent scheduled cardiac surgery between April 25th, 2022, and September 7th, 2023. Among them, 52 patients received sivelestat (administered at a dosage of 0.2m

The utilization of antibiotics is prevalent among lactating mothers. Hence, the rapid determination of trace amounts of antibiotics in human milk is crucial for ensuring the healthy development of infants. In this study, we constructed a human milk system containing residual doxycycline (DXC) and/or tetracycline (TC). Machine learning models and clustering algorithms were applied to classify and predict deficient concentrations of single and mixed antibiotics via label-free SERS spectra. The experimental results demonstrate that the CNN model has a recognition accuracy of 98.85% across optimal hyperparameter combinations. Furthermore, we employed Independent Component Analysis (ICA) and the pseudo-Siamese Convolutional Neural Network (pSCNN) to quantify the ratios of individual antibiotics in mixed human milk samples. Integrating the SERS technique with machine learning algorithms shows significant potential for rapid discrimination and precise quantification of single and mixed antibiotics at deficient concentrations in human milk.

Oleosomes are oil droplets abundant in oilseeds that can be extracted together with proteins for use in emulsion-like foods and can be used as natural, ready emulsions as alternatives to dairy and fat-based products. To make optimal use of the raw material, we look at simultaneous extraction of oleosomes and proteins from rapeseed. The combined approach of colloid milling and twin screw pressing resulted in a higher yield of oleosomes (70 % w/w) when compared to twin screw press alone (54 % w/w), which may be due to extensive cell wall disruption caused by colloid mill processing, as evidenced by SEM imaging. The protein yield remained constant irrespective of the degree of cell wall disruption (50 % w/w). In conclusion, combining the colloid mill and twin screw press is a straightforward way for the extraction of oleosome and proteins from rapeseeds for food application.

Considering the adverse effects of agricultural-food systems on both human health and the environment, this research aimed to identify sustainable diets, which are nutritious, culturally acceptable, affordable, and have low environmental impacts, based on self-reported diets in China. Dietary data was collected with a 3-day 24-h dietary recall and weight food record combined method among 10,324 subjects aged 18–64 year, who participated in the China Health Nutrition Survey 2011. Diet quality was assessed by the Chinese Healthy Eating Index 2016 (CHEI2016). Environmental impact was measured by greenhouse gas emissions (GHGE), total water use (TWU), and land use (LU), and diet costs were calculated using market prices of community surveys. Reduced rank regression derived dietary patterns with 34 food groups as predictor variables, and used CHEI2016 score, dietary greenhouse gas emissions (GHGE), total water use (TWU), land use (LU), and cost of the diet as response variables. Four dietary patterns were identified. Participants with the highest adherence (decile 10) to the “High animal-based food” pattern showed higher dietary GHGE (+57%), TWU (+51%), and LU (+54%) and dietary costs (+64%), compared to the average population's diets. The diet in decile 10 for the “High fruit, low ruminant meat” pattern displayed a 21% higher CHEI2016 score, and higher dietary environmental impact (GHGE +17%; TWU +22%; LU +19%) and dietary costs (+46%) than the average diets. Diets of participants who followed the “High fish, low beverages” pattern showed higher environmental impact (GHGE +39%; TWU +32%; LU +28%) and dietary costs (+19%), but the CHEI2016 score was similar (+0.1%). Finally, the “High wheat, low pork” pattern demonstrated lower environmental impacts (GHGE -17%, TWU -12%, LU -2%) and lower cost of the diet (−2%) but also lower CHEI2016 score (−1%) compared to average population. This study reveals the complex trade-offs between diet quality, environmental sustainability

Oleosomes are highly conserved organelles that store oil and can be utilized as a natural oil-in-water emulsion for food products when isolated intact. Although soybeans are a rich source of oleosomes, their potential has not been fully realized due to significant aggregation within the emulsion, which limits their applications. This study investigates the effect of homogenization as a strategy to prevent oleosome aggregation and coalescence. Our results indicate that homogenization at pressures of 10–30 MPa effectively prevents phase separation and preserves the initial particle size of oleosomes. However, at pressures of 40–50 MPa, oleosome particles are disrupted, leading to the formation of smaller particles that stabilize the emulsion against coalescence. We found that extracting oleosomes at a pH of 9 consistently produces stable emulsions, suggesting this method as a viable approach to enhance the stability of these emulsions. Furthermore, extraction at a pH of 7 followed by homogenization enhances the physical stability of the mixtures. Additionally, our findings demonstrate that performing extractions at pH 9 helps in preserving oleosome membrane-associated proteins, thereby further improving the stability and extending the storage potential of soybean oleosomes to up to 20 days.

Ethylene glycol (EG, 1,2-ethanediol) is a two-carbon dihydroxy alcohol that can be derived from fermentation of plant-derived xylose and arabinose and which can be formed during food fermentations. Here we show that Propionibacterium freudenreichii DSM 20271 is able to convert EG in anaerobic conditions to ethanol and acetate in almost equimolar amounts. The metabolism of EG led to a moderate increase of biomass, indicating its metabolism is energetically favourable. A proteomic analysis revealed EG induced expression of the pdu-cluster, which encodes a functional bacterial microcompartment (BMC) involved in the degradation of 1,2-propanediol, with the presence of BMCs confirmed using transmission electron microscopy. Cross-examination of the proteomes of 1,2-propanediol and EG grown cells revealed PDU BMC-expressing cells have elevated levels of DNA repair proteins and cysteine biosynthesis proteins. Cells grown in 1,2-propanediol and EG also showed enhanced resistance against acid and bile salt-induced stresses compared to lactate-grown cells. Our analysis of whole genome sequences of selected genomes of BMC-encoding microorganisms able to metabolize EG with acetaldehyde as intermediate indicate a potentially broad-distributed role of the pdu operon in metabolism of EG. Based on our results we conclude EG is metabolized to acetate and ethanol with acetaldehyde as intermediate within BMCs in P. freudenreichii.

Starting from holography for IIB string theory on AdS$_3\times \mathcal N$ with NS-NS flux, the TsT/$T\bar T$ correspondence is a conjecture that a TsT transformation on the string theory side is holographically dual to the single-trace version of the $T\bar T$ deformation on the field theory side. More precisely, the long string sector of string theory on the TsT-transformed background corresponds to the symmetric product theory whose seed theory is the $T\bar T$-deformed CFT$_2$. In this paper, we study the asymptotic symmetry of the string theory in the bulk. We find a state-dependent, non-local field redefinition under which the worldsheet equations of motion, stress tensor, as well as the symplectic form of string theory after the TsT transformation are mapped to those before the TsT transformation. The asymptotic symmetry in the auxiliary AdS basis is generated by two commuting Virasoro generators, while in the TsT transformed basis is non-linear and non-local. Our result from string theory analysis is compatible with that of the $T\bar T$ deformed CFT$_2$., Comment: 33 pages

Infectious diseases caused by microbial pathogens remain a primary contributor to global health burdens. Prompt control and effective prevention of these pathogens are critical for public health and medical diagnostics. Conventional microbial detection methods suffer from high complexity, low sensitivity, and poor selectivity. Therefore, developing rapid and reliable methods for microbial pathogen detection has become imperative. Surface-enhanced Raman Spectroscopy (SERS), as an innovative non-invasive diagnostic technique, holds significant promise in pathogenic microorganism detection due to its rapid, reliable, and cost-effective advantages. This review comprehensively outlines the fundamental theories of Raman Spectroscopy (RS) with a focus on label-free SERS strategy, reporting on the latest advancements of SERS technique in detecting bacteria, viruses, and fungi in clinical settings. Furthermore, we emphasize the application of machine learning algorithms in SERS spectral analysis. Finally, challenges faced by SERS application are probed, and the prospective development is discussed.

Recent research on housing submarkets over the decades and the necessity to categorise heterogeneous property markets into distinct submarkets are deemed essential in the real estate field. Housing submarkets operate based on the concept of a group of dwellings that are substantially close substitutes to one another, but are relatively poor substitutes for dwellings in different submarkets. This mechanism primarily depends on spatial variations in housing characteristics related to structure, location, and neighbourhood. Housing submarkets are vital in enhancing the precision of housing pricing and analysing the dynamic shifts in the property markets. This study aimed to define and examine the presence of housing submarkets for the terraced properties in Johor Bahru with the Malaysian Valuation and Property Services Department (JPPH) housing transaction database. The database documented 63,036 housing transactions between 2009 and 2018 with thirteen independent structural and locational housing attributes. The study also recommended an objective data-driven methodology by integrating principal component analysis (PCA) and cluster analysis to delineate housing submarkets. Resultantly, the presence of housing submarkets was identified within the property market with the delineation of two specific submarket effects, namely the housing location cluster and housing quality cluster. The submarkets provided a comprehensive framework for segregating property traits into prospective clusters. Establishing housing submarkets may provide valuable insights, implicit cluster details, and a more reliable hedonic price model prediction. The outcomes could be employed by property valuers, analysts, and urban planners for a sound comprehension of current real estate circumstances.

The chirality-flipping operators of light fermions are currently poorly constrained by experimental analyses due to the lack of interference with Standard Model (SM) amplitudes in traditional observables. In this work, we propose to investigate the semi-leptonic scalar/tensor four-fermion operators of electron and quarks through the transverse double spin asymmetry (DSA) at Electron-Ion Collider, where both the electron and proton beams could be highly transversely polarized. Due to the chirality-flipping nature of these operators, we demonstrate that their interference with the SM results in an unsuppressed contribution to the DSA, and could lead to non-trivial azimuthal $\cos2\phi$ and $\sin2\phi$ distributions that are linearly dependent on their Wilson coefficients. This new method has the potential to significantly improve the current constraints on these scalar/tensor four-fermion operators without relying on theoretical assumptions about other types of new physics effects, particularly for the tensor type operator of the $u$-quark. Additionally, our findings indicate that both the real and imaginary parts of these operators can be simultaneously constrained and offer a new opportunity for probing potential $CP$-violation effects. However, it is important to note that these results would be sensitive to the quark transversity distributions, which are currently poorly constrained by the experimental data, but could be significantly improved at the upcoming Electron-Ion Collider. Therefore, our work opens up a new avenue to utilize this new spin asymmetry for exploring the new physics effects from the scalar/tensor four-fermion operators., Comment: 8 pages, 4 figures, accepted version to appear in PRD

Aiming to evaluate the influence of microstructure on residual strength of hardmetals, conical indentation is implemented and validated as a simple and practical methodology to introduce controlled damage. Work is carried out in three fine-grained WC-Co cemented carbides with different binder content. In doing so, a contact damage map is first constructed through combined use of a wide range of indentation loads, detailed inspection by means of optical/electron microscopy and sequential tomography. Threshold indentation load values, defining the transition from deformation into cracking phenomena, are then used as experimental variable for residual strength testing of indented specimens. Dummy indentations are included in single specimens for extracting additional post-mortem information. Microstructural effects on residual strength are determined by comparing the fracture resistance under four-point bending of indented specimens with respect to values determined for pristine samples. It is found that fracture toughness prevails over hardness as key parameter for inducing a gradual strength lessening, instead of abrupt resistance drops. Main reason behind it is the capability of tougher hardmetal grades to uniformly distribute and homogenize the damage introduced during conical indentation. This is supported by the differences observed in the interaction of imprint and cracking features along failure paths as well as in stable growth of fissures linked to dummy indentations under the application of an external load. This finally translates in enhanced resistance to crack extension and higher damage tolerance., Peer Reviewed, Postprint (published version)

Antibiotic resistance genes (ARGs) are emerging pollutants that enter the farm and surrounding environment via the manure of antibiotic-treated animals. Pretreatment of livestock manure by composting decreases ARGs abundance, but how antibiotic residues affect ARGs removal efficiency remains poorly understood. Here, we explored the fate of the resistome under different doxycycline residue levels during aerobic swine manure composting. Metagenomic sequencing showed that the presence of high levels of doxycycline generally had a higher abundance of tetracycline ARGs, and their dominant host bacteria of Firmicutes, especially Clostridium and Streptococcus, also had limited elimination in composting under high levels of doxycycline stress. Moreover, high levels of doxycycline impaired the removal of the total ARGs number in finished composts, with a removal rate of 51.74 % compared to 63.70 % and 71.52 % for the control and low-level doxycycline manure, respectively. Horizontal gene transfer and strengthened correlations among the bacterial community fostered ARGs preservation at high doxycycline levels during composting. In addition, ARGs carried by both plasmids and chromosomes, such as multidrug ARGs, showed wide host characteristics and rebound during compost maturation. Compared with chromosomes, a greater variety of ARGs on plasmids suggested that the majority of ARGs were characterized by horizontal mobility during composting, and the cross-host characteristics of ARGs during composting deserve further attention. This study provided deep insight into the fate of ARGs under residual antibiotic stress during manure composting and reminded the associated risk for environmental and public health.

Foodborne illnesses, particularly those caused by Salmonella enterica with its extensive array of over 2600 serovars, present a significant public health challenge. Therefore, prompt and precise identification of S. enterica serovars is essential for clinical relevance, which facilitates the understanding of S. enterica transmission routes and the determination of outbreak sources. Classical serotyping methods via molecular subtyping and genomic markers currently suffer from various limitations, such as labour intensiveness, time consumption, etc. Therefore, there is a pressing need to develop new diagnostic techniques. Surface-enhanced Raman spectroscopy (SERS) is a non-invasive diagnostic technique that can generate Raman spectra, based on which rapid and accurate discrimination of bacterial pathogens could be achieved. To generate SERS spectra, a Raman spectrometer is needed to detect and collect signals, which are divided into two types: the expensive benchtop spectrometer and the inexpensive handheld spectrometer. In this study, we compared the performance of two Raman spectrometers to discriminate four closely associated S. enterica serovars, that is, S. enterica subsp. enterica serovar dublin, enteritidis, typhi and typhimurium. Six machine learning algorithms were applied to analyse these SERS spectra. The support vector machine (SVM) model showed the highest accuracy for both handheld (99.97%) and benchtop (99.38%) Raman spectrometers. This study demonstrated that handheld Raman spectrometers achieved similar prediction accuracy as benchtop spectrometers when combined with machine learning models, providing an effective solution for rapid, accurate and cost-effective identification of closely associated S. enterica serovars.

We investigate the suppression of star formation in galaxy pairs based on the isolated galaxy pair sample derived from the SDSS survey. By comparing the star formation rate between late-type galaxies in galaxy pairs and those in the isolated environment, we detect the signal of star formation suppression in galaxy pairs at $d_p < 100$kpc and $200$kpc$ < d_p < 350$kpc. The occurrence of star formation suppression in these late-type galaxies requires their companion galaxies to have an early-type morphology ($n_s > 2.5$). Star formation suppression in wide galaxy pairs with $200$kpc$ < d_p < 350$kpc mainly occurs in massive late-type galaxies, while in close galaxy pairs with $d_p < 100$kpc, it only appears in late-type galaxies with a massive companion ( $\log M_\star > 11.0$), nearly independent of their own stellar mass. Based on these findings, we infer that star formation suppression in wide galaxy pairs is actually a result of galaxy conformity, while in close galaxy pairs, it stems from the influence of hot circum-galactic medium surrounding companion galaxies., Comment: 10 pages, 4 figures, accepted for publication in ApJ

The spread of antibiotic resistance genes (ARGs) in agroecosystems through the application of animal manure is a global threat to human and environmental health. However, the adaptability and colonization ability of animal manure-derived bacteria determine the spread pathways of ARG in agroecosystems, which have rarely been studied. Here, we performed an invasion experiment by creating a synthetic communities (SynCom) with ten isolates from pig manure and followed its assembly during gnotobiotic cultivation of a soil-Arabidopsis thaliana (A. thaliana) system. We found that Firmicutes in the SynCom were efficiently filtered out in the rhizosphere, thereby limiting the entry of tetracycline resistance genes (TRGs) into the plant. However, Proteobacteria and Actinobacteria in the SynCom were able to establish in all compartments of the soil–plant system thereby spreading TRGs from manure to soil and plant. The presence of native soil bacteria prevented the establishment of manure-borne bacteria and effectively reduced the spread of TRGs. Achromobacter mucicolens and Pantoea septica were the main vectors for the entry of tetA into plants. Furthermore, doxycycline stress promoted the horizontal gene transfer (HGT) of the conjugative resistance plasmid RP4 within the SynCom in A. thaliana by upregulating the expression of HGT-related mRNAs. Therefore, this study provides evidence for the dissemination pathways of ARGs in agricultural systems through the invasion of manure-derived bacteria and HGT by conjugative resistance plasmids and demonstrates that the priority establishment of soil bacteria in the rhizosphere limited the spread of TRGs from pig manure to soil-plant systems.

Yunpeng Feng,1 Shifeng Qiao,1 Jie Chen,1 Xin Wen,2 Yanlei Chen,3 Xiaoyu Song,1 Jiaxin Xu,1 Xiucheng Qiao,1 Jing Yang,2 Shenshen Zhang,1 Yang Feng,1 Yu Gao4 1The Second Ward of Colorectal Surgery, The First Affiliated Hospital of Jinzhou Medical University, Jinzhou, Liaoning, 12100, People’s Republic of China; 2Department of Pathology, The First Affiliated Hospital of Jinzhou Medical University, Jinzhou, Liaoning, 12100, People’s Republic of China; 3The First Affiliated Hospital of Sun Yat-sen University, Guangzhou, Guangdong, People’s Republic of China; 4School of Basic Medical Sciences, Jinzhou Medical University, Jinzhou, Liaoning, 12100, People’s Republic of ChinaCorrespondence: Yu Gao, School of Basic Medical Sciences, Jinzhou Medical University, No. 40, Section 3 Songpo Road, Jinzhou, Liaoning, 121000, People’s Republic of China, Tel +86 15904161717, Email jzykdxgy@163.comPurpose: This research explored the association between CD163-labeled M2-type macrophages and cancer-associated fibroblasts (CAFs) in the tumor microenvironment (TME) of 38 colorectal cancer (CRC) liver metastases. In addition, we investigated the correlation differences between M2-type macrophages and CAFs in the tumor microenvironments of 38 primary colorectal cancer patients with confirmed liver metastases and 946 colorectal cancer patients, as well as possible mechanisms of action between the two cells.Methods: The Immunohistochemistry (IHC) method was applied to detect the expression levels of M2-type macrophages and CAFs in the tissues of 984 cases of CRC and to analyze the correlation between M2-type macrophages and CAFs in colorectal cancer tissues. The IHC method was also applied to detect the expression levels of M2-type macrophages and CAFs in the liver metastases of 38 cases of CRC in the experimental group and to analyze the correlation between the two cells in liver metastases.Results: 1. M2-type macrophages and CAFs expression were significantly higher in 38 primary co

The spread of livestock manure-borne antibiotic resistance genes (ARGs) into agroecosystems through manure application poses a potential threat to human health. However, there is still a knowledge gap concerning ARG dissemination in coalescing manure, soil and plant microbiomes. Here, we examined the fate of tetracycline resistance genes (TRGs) originating from pig manure microbiomes and spread in the soil-A thaliana system and explored the effects of microbial functions on TRGs spread at different interfaces. Our results indicate that the TRGs abundances in all microbiome continuum of the soil-A. thaliana system were significantly increased with the application of a living manure microbiome, although the addition of manure with both an active and inactive microbiome caused a shift in the microbial community composition. This was attributed to the increasing relative abundances of tetA, tetL, tetM, tetO, tetW and tolC in the system. The application of living manure with DOX residues resulted in the highest relative abundance of total TRGs (3.30×10−3 copies/16S rRNA gene copies) in the rhizosphere soil samples. Community coalescence of the manure and soil microbiomes increased the abundance of Firmicutes in the soil and root microbiome, which directly explains the increase in TRG abundance observed in these interfaces. In contrast, the leaf microbiome differed markedly from that of the remaining samples, indicating strong plant host filtering effects on Firmicutes and TRGs from pig manure. The random forest machine learning model revealed microbial functions and their significant positive correlation with TRG abundance in the microbiome continuum of the system. Our findings revealed that community coalescence is the main driver of TRG spread from manure to the soil and root microbiomes. Plant host filtering effects play a crucial role in allowing certain microbial groups to occupy ecological niches in the leaves, thereby limiting the establishment of manure-borne TRG

Recently, there has been a growing interest in automatic software vulnerability detection. Pre-trained model-based approaches have demonstrated superior performance than other Deep Learning (DL)-based approaches in detecting vulnerabilities. However, the existing pre-trained model-based approaches generally employ code sequences as input during prediction, and may ignore vulnerability-related structural information, as reflected in the following two aspects. First, they tend to fail to infer the semantics of the code statements with complex logic such as those containing multiple operators and pointers. Second, they are hard to comprehend various code execution sequences, which is essential for precise vulnerability detection. To mitigate the challenges, we propose a Structured Natural Language Comment tree-based vulnerAbiLity dEtection framework based on the pre-trained models, named SCALE. The proposed Structured Natural Language Comment Tree (SCT) integrates the semantics of code statements with code execution sequences based on the Abstract Syntax Trees (ASTs). Specifically, SCALE comprises three main modules: (1) Comment Tree Construction, which aims at enhancing the model's ability to infer the semantics of code statements by first incorporating Large Language Models (LLMs) for comment generation and then adding the comment node to ASTs. (2) Structured Natural Language Comment Tree Construction}, which aims at explicitly involving code execution sequence by combining the code syntax templates with the comment tree. (3) SCT-Enhanced Representation, which finally incorporates the constructed SCTs for well capturing vulnerability patterns., Comment: Accepted by ISSTA 2024

In this paper, we present an empirical study on image recognition fairness, i.e., extreme class accuracy disparity on balanced data like ImageNet. We experimentally demonstrate that classes are not equal and the fairness issue is prevalent for image classification models across various datasets, network architectures, and model capacities. Moreover, several intriguing properties of fairness are identified. First, the unfairness lies in problematic representation rather than classifier bias. Second, with the proposed concept of Model Prediction Bias, we investigate the origins of problematic representation during optimization. Our findings reveal that models tend to exhibit greater prediction biases for classes that are more challenging to recognize. It means that more other classes will be confused with harder classes. Then the False Positives (FPs) will dominate the learning in optimization, thus leading to their poor accuracy. Further, we conclude that data augmentation and representation learning algorithms improve overall performance by promoting fairness to some degree in image classification. The Code is available at https://github.com/dvlab-research/Parametric-Contrastive-Learning., Comment: CVPR 2024

This paper addresses the challenge of integrating sequentially arriving data within the quantile regression framework, where the number of features is allowed to grow with the number of observations, the horizon is unknown, and memory is limited. We employ stochastic sub-gradient descent to minimize the empirical check loss and study its statistical properties and regret performance. In our analysis, we unveil the delicate interplay between updating iterates based on individual observations versus batches of observations, revealing distinct regularity properties in each scenario. Our method ensures long-term optimal estimation irrespective of the chosen update strategy. Importantly, our contributions go beyond prior works by achieving exponential-type concentration inequalities and attaining optimal regret and error rates that exhibit only \textsf{ short-term} sensitivity to initial errors. A key insight from our study is the delicate statistical analyses and the revelation that appropriate stepsize schemes significantly mitigate the impact of initial errors on subsequent errors and regrets. This underscores the robustness of stochastic sub-gradient descent in handling initial uncertainties, emphasizing its efficacy in scenarios where the sequential arrival of data introduces uncertainties regarding both the horizon and the total number of observations. Additionally, when the initial error rate is well-controlled, there is a trade-off between short-term error rate and long-term optimality. Due to the lack of delicate statistical analysis for squared loss, we also briefly discuss its properties and proper schemes. Extensive simulations support our theoretical findings.

Open-Source Software (OSS) vulnerabilities bring great challenges to the software security and pose potential risks to our society. Enormous efforts have been devoted into automated vulnerability detection, among which deep learning (DL)-based approaches have proven to be the most effective. However, the current labeled data present the following limitations: (1) Tangled Patches: Developers may submit code changes unrelated to vulnerability fixes within patches, leading to tangled patches. (2) Lacking Inter-procedural Vulnerabilities: The existing vulnerability datasets typically contain function-level and file-level vulnerabilities, ignoring the relations between functions, thus rendering the approaches unable to detect the inter-procedural vulnerabilities. (3) Outdated Patches: The existing datasets usually contain outdated patches, which may bias the model during training. To address the above limitations, in this paper, we propose an automated data collection framework and construct the first repository-level high-quality vulnerability dataset named ReposVul. The proposed framework mainly contains three modules: (1) A vulnerability untangling module, aiming at distinguishing vulnerability-fixing related code changes from tangled patches, in which the Large Language Models (LLMs) and static analysis tools are jointly employed. (2) A multi-granularity dependency extraction module, aiming at capturing the inter-procedural call relationships of vulnerabilities, in which we construct multiple-granularity information for each vulnerability patch, including repository-level, file-level, function-level, and line-level. (3) A trace-based filtering module, aiming at filtering the outdated patches, which leverages the file path trace-based filter and commit time trace-based filter to construct an up-to-date dataset., Comment: Accepted by ICSE 2024 Industry Challenge Track

Recent years have witnessed a growing focus on automated software vulnerability detection. Notably, deep learning (DL)-based methods, which employ source code for the implicit acquisition of vulnerability patterns, have demonstrated superior performance compared to other approaches. However, the DL-based approaches are still hard to capture the vulnerability-related information from the whole code snippet, since the vulnerable parts usually account for only a small proportion. As evidenced by our experiments, the approaches tend to excessively emphasize semantic information, potentially leading to limited vulnerability detection performance in practical scenarios. First, they cannot well distinguish between the code snippets before (i.e., vulnerable code) and after (i.e., non-vulnerable code) developers' fixes due to the minimal code changes. Besides, substituting user-defined identifiers with placeholders (e.g., "VAR1" and "FUN1") in obvious performance degradation at up to 14.53% with respect to the F1 score. To mitigate these issues, we propose to leverage the vulnerable and corresponding fixed code snippets, in which the minimal changes can provide hints about semantic-agnostic features for vulnerability detection. In this paper, we propose a software vulneRability dEteCtion framework with zerO-sum game and prototype learNing, named RECON. In RECON, we propose a zero-sum game construction module. Distinguishing the vulnerable code from the corresponding fixed code is regarded as one player (i.e. Calibrator), while the conventional vulnerability detection is another player (i.e. Detector) in the zero-sum game. The goal is to capture the semantic-agnostic features of the first player for enhancing the second player's performance for vulnerability detection. Experiments on the public benchmark dataset show that RECON outperforms the state-of-the-art baseline by 6.29% in F1 score., Comment: 17 pages, 8 figures

Severe data imbalance naturally exists among web-scale vision-language datasets. Despite this, we find CLIP pre-trained thereupon exhibits notable robustness to the data imbalance compared to supervised learning, and demonstrates significant effectiveness in learning generalizable representations. With an aim to investigate the reasons behind this finding, we conduct controlled experiments to study various underlying factors, and reveal that CLIP's pretext task forms a dynamic classification problem wherein only a subset of classes is present in training. This isolates the bias from dominant classes and implicitly balances the learning signal. Furthermore, the robustness and discriminability of CLIP improve with more descriptive language supervision, larger data scale, and broader open-world concepts, which are inaccessible to supervised learning. Our study not only uncovers the mechanisms behind CLIP's generalizability beyond data imbalance but also provides transferable insights for the research community. The findings are validated in both supervised and self-supervised learning, enabling models trained on imbalanced data to achieve CLIP-level performance on diverse recognition tasks. Code will be available at: https://github.com/CVMI-Lab/clip-beyond-tail.

The data-driven newsvendor problem with features has recently emerged as a significant area of research, driven by the proliferation of data across various sectors such as retail, supply chains, e-commerce, and healthcare. Given the sensitive nature of customer or organizational data often used in feature-based analysis, it is crucial to ensure individual privacy to uphold trust and confidence. Despite its importance, privacy preservation in the context of inventory planning remains unexplored. A key challenge is the nonsmoothness of the newsvendor loss function, which sets it apart from existing work on privacy-preserving algorithms in other settings. This paper introduces a novel approach to estimate a privacy-preserving optimal inventory policy within the f-differential privacy framework, an extension of the classical $(\epsilon, \delta)$-differential privacy with several appealing properties. We develop a clipped noisy gradient descent algorithm based on convolution smoothing for optimal inventory estimation to simultaneously address three main challenges: (1) unknown demand distribution and nonsmooth loss function; (2) provable privacy guarantees for individual-level data; and (3) desirable statistical precision. We derive finite-sample high-probability bounds for optimal policy parameter estimation and regret analysis. By leveraging the structure of the newsvendor problem, we attain a faster excess population risk bound compared to that obtained from an indiscriminate application of existing results for general nonsmooth convex loss. Our bound aligns with that for strongly convex and smooth loss function. Our numerical experiments demonstrate that the proposed new method can achieve desirable privacy protection with a marginal increase in cost.

Deep Learning (DL)-based methods have proven to be effective for software vulnerability detection, with a potential for substantial productivity enhancements for detecting vulnerabilities. Current methods mainly focus on detecting single functions (i.e., intra-procedural vulnerabilities), ignoring the more complex inter-procedural vulnerability detection scenarios in practice. For example, developers routinely engage with program analysis to detect vulnerabilities that span multiple functions within repositories. In addition, the widely-used benchmark datasets generally contain only intra-procedural vulnerabilities, leaving the assessment of inter-procedural vulnerability detection capabilities unexplored. To mitigate the issues, we propose a repository-level evaluation system, named \textbf{VulEval}, aiming at evaluating the detection performance of inter- and intra-procedural vulnerabilities simultaneously. Specifically, VulEval consists of three interconnected evaluation tasks: \textbf{(1) Function-Level Vulnerability Detection}, aiming at detecting intra-procedural vulnerability given a code snippet; \textbf{(2) Vulnerability-Related Dependency Prediction}, aiming at retrieving the most relevant dependencies from call graphs for providing developers with explanations about the vulnerabilities; and \textbf{(3) Repository-Level Vulnerability Detection}, aiming at detecting inter-procedural vulnerabilities by combining with the dependencies identified in the second task. VulEval also consists of a large-scale dataset, with a total of 4,196 CVE entries, 232,239 functions, and corresponding 4,699 repository-level source code in C/C++ programming languages. Our analysis highlights the current progress and future directions for software vulnerability detection., Comment: 12 pages

Preventing spontaneous crystallization of supersaturated solutions by additives is of critical interest to successful process design and implementation, with numerous applications in chemical, pharmaceutical, medical, pigment, and food industries, but challenges remain in laboratory and industry settings and fundamental understanding is lacking. When copresented with antifreeze proteins (AFPs), otherwise spontaneously crystallizing osmolytes are maintained at high supersaturations for months in over-wintering organisms. Thus, we here explore the inhibition phenomenon by AFPs, using persistent crystallization of a common sugar alcohol, D-mannitol, as a case study. We report experimentally that DAFP1, an insect AFP, completely inhibits D-mannitol nucleation. Computer simulations reveal a new mechanism for crystallization inhibition where the population of the crystal-forming conformers are selectively bound and randomized in solution by hydrogen bonding to the protein surface. These results highlight the advantages of using natural polymers to address crystallization inhibition challenges and suggest new strategies in controlling the nucleation processes.

$T\bar T$ deformed CFTs with positive deformation parameter have been proposed to be holographically dual to Einstein gravity in a glue-on $\mathrm{AdS}_3$ spacetime. The latter is constructed from AdS$_3$ by gluing a patch of an auxiliary AdS$_3^*$ spacetime to its asymptotic boundary. In this work, we propose a glue-on version of the Ryu-Takayanagi formula, which is given by the signed area of an extremal surface. The extremal surface is anchored at the endpoints of an interval on a cutoff surface in the glue-on geometry. It consists of an RT surface lying in the AdS$_3$ part of the spacetime and its extension to the AdS$_3^*$ region. The signed area is the length of the RT surface minus the length of the segments in AdS$_3^*$. We find that the Ryu-Takayanagi formula with the signed area reproduces the entanglement entropy of a half interval for $T\bar T$-deformed CFTs on the sphere. We then study the properties of extremal surfaces on various glue-on geometries, including Poincar\'e $\mathrm{AdS}_3$, global $\mathrm{AdS}_3$, and the BTZ black hole. When anchored on multiple intervals at the boundary, the signed area of the minimal surfaces undergoes phase transitions with novel properties. In all of these examples, we find that the glue-on extremal surfaces exhibit a minimum length related to the deformation parameter of $T\bar T$-deformed CFTs., Comment: 39 pages, 9 figures; v2: added references; v3: fixed typos, matches published version

The $T\bar T$ deformation is a solvable irrelevant deformation whose properties depend on the sign of the deformation parameter $\mu$. In particular, $T\bar T$-deformed CFTs with $\mu<0$ have been proposed to be holographically dual to Einstein gravity where the metric satisfies Dirichlet boundary conditions at a finite cutoff surface. In this paper, we put forward a holographic proposal for $T\bar T$-deformed CFTs with $\mu>0$, in which case the bulk geometry is constructed by gluing a patch of AdS$_3$ to the original spacetime. As evidence, we show that the $T\bar T$ trace flow equation, the spectrum on the cylinder, and the partition function on the torus and the sphere, among other results, can all be reproduced from bulk calculations in glue-on AdS$_3$., Comment: 33 pages, 1 figure; v2: clarifications and references added, matches published version

Electroweak dipole operators in the Standard Model Effective Field Theory (SMEFT) are important indirect probes of quantum effects of new physics beyond the Standard Model (SM), yet they remain poorly constrained by current experimental analyses for lack of interference with the SM amplitudes in constructing cross section observables. In this Letter, we point out that dipole operators flip fermion helicities so are ideally studied through single transverse spin asymmetries. We illustrate this at a future electron-positron collider with transversely polarized beams, where such effect exhibits as azimuthal $\cos\phi$ and $\sin\phi$ distributions which originate from the interference of the electron dipole operators with the SM and are linearly dependent on their Wilson coefficients. This new method can improve the current constraints on the electron dipole couplings by one to two orders of magnitude, without depending on other new physics operators, and can also simultaneously constrain both their real and imaginary parts, offering a new opportunity for probing potential $CP$-violating effects., Comment: 5 pages, 1 figure

The thermodynamics of finite size quark matter in the quasiparticle model is self-consistently constructed by an effective bag function, which presents the medium effect to the confinement. We obtained completely analytic surface tension in the strong magnetic field with the multiple reflection expansion. The anisotropic structure is demonstrated by the splitting of the longitudinal and transverse surface tensions. The anisotropy of the surface tension could be enhanced by an increase of the magnetic field. The analytical surface tension is modified by an additional term related to the bag function. For strong enough magnetic fields, the increase of the longitudinal surface tension is proportional to the magnetic field.On the contrary, the transverse component vanishes due to all quarks locating in the lowest landau level., Comment: 9 pages, 5 figures

Hollow Ag–Pd nanoparticles have potentially high catalytic performance owing to their larger surface area compared to their corresponding solid nanoparticles. We successfully fabricated hollow Ag–Pd alloy nanodendrites and nanoboxes by using different Pd precursors (H2PdCl4 and Pd(acac)2) to achieve large surface area nanoboxes. Interestingly, the use of a H2PdCl4 precursor led to the formation of hollow nanodendrite structures, whereas the slower reduction of Pd(acac)2 led to the formation of hollow nanoboxes. The microstructure and chemical composition of Ag–Pd nanoparticles and properties of their growth solutions were investigated by transmission electron microscopy, energy-dispersive X-ray spectroscopy, and ultraviolet–visible spectroscopy., This work is supported from the Knut and Alice Wallenberg Foundation and the Swedish Research Council., Peer Reviewed, Postprint (published version)

Background The study employs a five-stage methodology to investigate the symbiotic potential of the Sciffle Box, human values, and value communication through gaming. This journey involves a literature review focused on understanding values and corporate values, along with research on the Sciffle Box and its contextual use. Three co-design sessions are conducted with master students at Delft University of Technology, identifying various contexts using a value based Sciffle Box game. Insights are synthesized into an opportunity map, followed by iterative design of five prototypes, culminating in the deployment of the most effective design. The final prototype is built with Unity on the Sciffle Box, evaluated by seven staff of Ijsfontein. The final design mapped out five different working dilemmas that are merged in five narratives presented by non-game players played by ChatGPT. Each dilemma has a behaviour goal which matches the corporate values. Players are divided into four teams; each team will have five different human values cards that can only be used once. By applying those values c, Design for Interaction

Hollow Ag-Pd nanoparticles have potentially high catalytic performance owing to their larger surface area compared to their corresponding solid nanoparticles. We successfully fabricated hollow Ag-Pd alloy nanodendrites and nanoboxes by using different Pd precursors (H2PdCl4 and Pd(acac)2) to achieve large surface area nanoboxes. Interestingly, the use of a H2PdCl4 precursor led to the formation of hollow nanodendrite structures, whereas the slower reduction of Pd(acac)2 led to the formation of hollow nanoboxes. The microstructure and chemical composition of Ag-Pd nanoparticles and properties of their growth solutions were investigated by transmission electron microscopy, energy-dispersive X-ray spectroscopy, and ultraviolet-visible spectroscopy.

Background: Increased urbanization has been linked to transitions in dietary patterns. However, evidence on the impacts of urbanization on diet quality, and environmental impact, and diet cost is limited. The aim of this study was to investigate the time trends of these three dietary sustainability in China over the period 1997–2011 and to examine their associations with urbanization. Methods: Food consumption of 8,330 participants (18–64y) of the China Health and Nutrition Survey cohort (1997, 2000, 2004, 2006, 2009, and 2011) were examined and diet quality was assessed using the Chinese Healthy Eating Index 2016 (CHEI2016). Dietary related environmental impacts on Greenhouse Gas Emissions (GHGE), Total Water Use (TWU), and Land Use (LU) were estimated using the Chinese Food Life Cycle Assessment Database. Monetary cost of diet was calculated using the community market prices of food items. Multilevel mixed-effects models were used to estimate associations between the time trend of dietary sustainability indicators and degree of urbanization. Results: From 1997 to 2011, the CHEI2016 score increased by 10.6%, GHGE by 23.8%, LU by 29.1%, and the inflation-corrected cost of diet by 80%. Urbanization was positively associated with these time trends, which remained after adjustment for sociodemographic and lifestyle factors (all P < 0·05). Conclusion: The rapid urbanization in China over the past two decades has been followed by an improvement in the overall dietary quality, but this has been accompanied by an increase in the environmental impacts and higher cost of the diet, especially in communities with lower urbanization index.

With the Schwinger's proper-time formalism of the Nambu-Jona-Lasinio model, we investigate the finite volume effects in the presence of magnetic fields. Since the coupling constant $G$ can be influenced by strong magnetic fields, the model is solved with a running coupling constant $G(B)$ which is fitted by the lattice average $(\Sigma_u+\Sigma_d)/2$ and difference $\Sigma_u-\Sigma_d$. The investigation mainly focuses on the constituent quark mass and the thermal susceptibility depending on the magnetic fields, the temperatures and the finite sizes. For the model in finite or infinite volume, the magnetic fields can increase the constituent quark mass while the temperatures can decrease it inversely. There is a narrow range of the box length that makes the effects of finite volume perform prominently. The model will behave close to infinite volume limit for larger box length. It is shown that the influence of finite volume can be changed by magnetic fields and temperatures. Finally, we discuss the thermal susceptibility depending on the temperature in finite volume in the presence of magnetic fields., Comment: 13 pages, 6 figures

Due to the strong correlation between adjacent pixels, most image encryption schemes perform multiple rounds of confusion and diffusion to protect the image against attacks. Such operations, however, are time-consuming, cannot meet the real-time requirements of video encryption. Existing works, therefore, realize video encryption by simplifying the encryption process or encrypting specific parts of video frames, which results in lower security compared to image encryption. To solve the problem, this paper proposes a real-time chaotic video encryption strategy based on multithreaded parallel confusion and diffusion. It takes a video as the input, splits the frame into subframes, creates a set of threads to simultaneously perform five rounds of confusion and diffusion operations on corresponding subframes, and efficiently outputs the encrypted frames. The encryption speed evaluation shows that our method significantly improves the confusion and diffusion speed, realizes real-time 480x480, 576x576, and 768x768 24FPS video encryption using Intel Core i5-1135G7, Intel Core i7-8700, and Intel Xeon Gold 6226R, respectively. The statistical and security analysis prove that the deployed cryptosystems have outstanding statistical properties, can resist attacks, channel noise, and data loss. Compared with existing works, to the best of our knowledge, the proposed strategy achieves the fastest encryption speed, and realizes the first real-time chaotic video encryption that reaches the security level of image encryption. In addition, it is suitable for many confusion, diffusion algorithms and can be easily deployed with both hardware and software., Comment: 14 pages, 11 figures, 9 tables

Prior studies have demonstrated the effectiveness of Deep Learning (DL) in automated software vulnerability detection. Graph Neural Networks (GNNs) have proven effective in learning the graph representations of source code and are commonly adopted by existing DL-based vulnerability detection methods. However, the existing methods are still limited by the fact that GNNs are essentially difficult to handle the connections between long-distance nodes in a code structure graph. Besides, they do not well exploit the multiple types of edges in a code structure graph (such as edges representing data flow and control flow). Consequently, despite achieving state-of-the-art performance, the existing GNN-based methods tend to fail to capture global information (i.e., long-range dependencies among nodes) of code graphs. To mitigate these issues, in this paper, we propose a novel vulnerability detection framework with grAph siMplification and enhanced graph rePresentation LEarning, named AMPLE. AMPLE mainly contains two parts: 1) graph simplification, which aims at reducing the distances between nodes by shrinking the node sizes of code structure graphs; 2) enhanced graph representation learning, which involves one edge-aware graph convolutional network module for fusing heterogeneous edge information into node representations and one kernel-scaled representation module for well capturing the relations between distant graph nodes. Experiments on three public benchmark datasets show that AMPLE outperforms the state-of-the-art methods by 0.39%-35.32% and 7.64%-199.81% with respect to the accuracy and F1 score metrics, respectively. The results demonstrate the effectiveness of AMPLE in learning global information of code graphs for vulnerability detection., Comment: 13 pages, 8 figures, Accepted for publication in the ICSE 23 Technical Track

In this paper, we address the problem of generalized category discovery (GCD), \ie, given a set of images where part of them are labelled and the rest are not, the task is to automatically cluster the images in the unlabelled data, leveraging the information from the labelled data, while the unlabelled data contain images from the labelled classes and also new ones. GCD is similar to semi-supervised learning (SSL) but is more realistic and challenging, as SSL assumes all the unlabelled images are from the same classes as the labelled ones. We also do not assume the class number in the unlabelled data is known a-priori, making the GCD problem even harder. To tackle the problem of GCD without knowing the class number, we propose an EM-like framework that alternates between representation learning and class number estimation. We propose a semi-supervised variant of the Gaussian Mixture Model (GMM) with a stochastic splitting and merging mechanism to dynamically determine the prototypes by examining the cluster compactness and separability. With these prototypes, we leverage prototypical contrastive learning for representation learning on the partially labelled data subject to the constraints imposed by the labelled data. Our framework alternates between these two steps until convergence. The cluster assignment for an unlabelled instance can then be retrieved by identifying its nearest prototype. We comprehensively evaluate our framework on both generic image classification datasets and challenging fine-grained object recognition datasets, achieving state-of-the-art performance., Comment: This paper is accepted at ICCV 2023

Pre-trained models of source code have gained widespread popularity in many code intelligence tasks. Recently, with the scaling of the model and corpus size, large language models have shown the ability of in-context learning (ICL). ICL employs task instructions and a few examples as demonstrations, and then inputs the demonstrations to the language models for making predictions. This new learning paradigm is training-free and has shown impressive performance in various natural language processing and code intelligence tasks. However, the performance of ICL heavily relies on the quality of demonstrations, e.g., the selected examples. It is important to systematically investigate how to construct a good demonstration for code-related tasks. In this paper, we empirically explore the impact of three key factors on the performance of ICL in code intelligence tasks: the selection, order, and number of demonstration examples. We conduct extensive experiments on three code intelligence tasks including code summarization, bug fixing, and program synthesis. Our experimental results demonstrate that all the above three factors dramatically impact the performance of ICL in code intelligence tasks. Additionally, we summarize our findings and provide takeaway suggestions on how to construct effective demonstrations, taking into account these three perspectives. We also show that a carefully-designed demonstration based on our findings can lead to substantial improvements over widely-used demonstration construction methods, e.g., improving BLEU-4, EM, and EM by at least 9.90%, 175.96%, and 50.81% on code summarization, bug fixing, and program synthesis, respectively, Comment: This paper is accepted by ASE 2023

As a pioneering work, PointContrast conducts unsupervised 3D representation learning via leveraging contrastive learning over raw RGB-D frames and proves its effectiveness on various downstream tasks. However, the trend of large-scale unsupervised learning in 3D has yet to emerge due to two stumbling blocks: the inefficiency of matching RGB-D frames as contrastive views and the annoying mode collapse phenomenon mentioned in previous works. Turning the two stumbling blocks into empirical stepping stones, we first propose an efficient and effective contrastive learning framework, which generates contrastive views directly on scene-level point clouds by a well-curated data augmentation pipeline and a practical view mixing strategy. Second, we introduce reconstructive learning on the contrastive learning framework with an exquisite design of contrastive cross masks, which targets the reconstruction of point color and surfel normal. Our Masked Scene Contrast (MSC) framework is capable of extracting comprehensive 3D representations more efficiently and effectively. It accelerates the pre-training procedure by at least 3x and still achieves an uncompromised performance compared with previous work. Besides, MSC also enables large-scale 3D pre-training across multiple datasets, which further boosts the performance and achieves state-of-the-art fine-tuning results on several downstream tasks, e.g., 75.5% mIoU on ScanNet semantic segmentation validation set., Comment: Accepted by CVPR 2023, available at https://github.com/Pointcept/Pointcept

Recent technological advances have led to contemporary applications that demand real-time processing and analysis of sequentially arriving tensor data. Traditional offline learning, involving the storage and utilization of all data in each computational iteration, becomes impractical for high-dimensional tensor data due to its voluminous size. Furthermore, existing low-rank tensor methods lack the capability for statistical inference in an online fashion, which is essential for real-time predictions and informed decision-making. This paper addresses these challenges by introducing a novel online inference framework for low-rank tensor learning. Our approach employs Stochastic Gradient Descent (SGD) to enable efficient real-time data processing without extensive memory requirements, thereby significantly reducing computational demands. We establish a non-asymptotic convergence result for the online low-rank SGD estimator, nearly matches the minimax optimal rate of estimation error in offline models that store all historical data. Building upon this foundation, we propose a simple yet powerful online debiasing approach for sequential statistical inference in low-rank tensor learning. The entire online procedure, covering both estimation and inference, eliminates the need for data splitting or storing historical data, making it suitable for on-the-fly hypothesis testing. Given the sequential nature of our data collection, traditional analyses relying on offline methods and sample splitting are inadequate. In our analysis, we control the sum of constructed super-martingales to ensure estimates along the entire solution path remain within the benign region. Additionally, a novel spectral representation tool is employed to address statistical dependencies among iterative estimates, establishing the desired asymptotic normality.

Artificial Intelligence(AI) widely applies in Image Classification and Recognition, Text Understanding and Natural Language Processing, which makes great progress. In this paper, we introduced AI into the fruit quality detection field. We designed a fruit sugar degree regression model using an Artificial Neural Network based on spectra of fruits within the visible/near-infrared(V/NIR)range. After analysis of fruit spectra, we innovatively proposed a new neural network structure: low layers consist of a Multilayer Perceptron(MLP), a middle layer is a 2-dimensional correlation matrix layer, and high layers consist of several Convolutional Neural Network(CNN) layers. In this study, we used fruit sugar value as a detection target, collecting two fruits called Gan Nan Navel and Tian Shan Pear as samples, doing experiments respectively, and comparing their results. We used Analysis of Variance(ANOVA) to evaluate the reliability of the dataset we collected. Then, we tried multiple strategies to process spectrum data, evaluating their effects. In this paper, we tried to add Wavelet Decomposition(WD) to reduce feature dimensions and a Genetic Algorithm(GA) to find excellent features. Then, we compared Neural Network models with traditional Partial Least Squares(PLS) based models. We also compared the neural network structure we designed(MLP-CNN) with other traditional neural network structures. In this paper, we proposed a new evaluation standard derived from dataset standard deviation(STD) for evaluating detection performance, validating the viability of using an artificial neural network model to do fruit sugar degree nondestructive detection.

Deriving reliable region-word alignment from image-text pairs is critical to learn object-level vision-language representations for open-vocabulary object detection. Existing methods typically rely on pre-trained or self-trained vision-language models for alignment, which are prone to limitations in localization accuracy or generalization capabilities. In this paper, we propose CoDet, a novel approach that overcomes the reliance on pre-aligned vision-language space by reformulating region-word alignment as a co-occurring object discovery problem. Intuitively, by grouping images that mention a shared concept in their captions, objects corresponding to the shared concept shall exhibit high co-occurrence among the group. CoDet then leverages visual similarities to discover the co-occurring objects and align them with the shared concept. Extensive experiments demonstrate that CoDet has superior performances and compelling scalability in open-vocabulary detection, e.g., by scaling up the visual backbone, CoDet achieves 37.0 $\text{AP}^m_{novel}$ and 44.7 $\text{AP}^m_{all}$ on OV-LVIS, surpassing the previous SoTA by 4.2 $\text{AP}^m_{novel}$ and 9.8 $\text{AP}^m_{all}$. Code is available at https://github.com/CVMI-Lab/CoDet., Comment: Accepted by NeurIPS 2023

Counterfactual reasoning, a fundamental aspect of human cognition, involves contemplating alternatives to established facts or past events, significantly enhancing our abilities in planning and decision-making. In light of the advancements in current multi-modal large language models, we explore their effectiveness in counterfactual reasoning. To facilitate this investigation, we introduce a novel dataset, C-VQA, specifically designed to test the counterfactual reasoning capabilities of modern multi-modal large language models. This dataset is constructed by infusing original questions with counterfactual presuppositions, spanning various types such as numerical and boolean queries. It encompasses a mix of real and synthetic data, representing a wide range of difficulty levels. Our thorough evaluations of contemporary vision-language models using this dataset have revealed substantial performance drops, with some models showing up to a 40% decrease, highlighting a significant gap between current models and human-like vision reasoning capabilities. We hope our dataset will serve as a vital benchmark for evaluating the counterfactual reasoning capabilities of models. Code and dataset are publicly available at https://bzhao.me/C-VQA/.

The rapid advancement of deep learning models often attributes to their ability to leverage massive training data. In contrast, such privilege has not yet fully benefited 3D deep learning, mainly due to the limited availability of large-scale 3D datasets. Merging multiple available data sources and letting them collaboratively train a single model is a potential solution. However, due to the large domain gap between 3D point cloud datasets, such mixed supervision could adversely affect the model's performance and lead to degenerated performance (i.e., negative transfer) compared to single-dataset training. In view of this challenge, we introduce Point Prompt Training (PPT), a novel framework for multi-dataset synergistic learning in the context of 3D representation learning that supports multiple pre-training paradigms. Based on this framework, we propose Prompt-driven Normalization, which adapts the model to different datasets with domain-specific prompts and Language-guided Categorical Alignment that decently unifies the multiple-dataset label spaces by leveraging the relationship between label text. Extensive experiments verify that PPT can overcome the negative transfer associated with synergistic learning and produce generalizable representations. Notably, it achieves state-of-the-art performance on each dataset using a single weight-shared model with supervised multi-dataset training. Moreover, when served as a pre-training framework, it outperforms other pre-training approaches regarding representation quality and attains remarkable state-of-the-art performance across over ten diverse downstream tasks spanning both indoor and outdoor 3D scenarios., Comment: Code available at Pointcept (https://github.com/Pointcept/Pointcept)

Automated code vulnerability detection has gained increasing attention in recent years. The deep learning (DL)-based methods, which implicitly learn vulnerable code patterns, have proven effective in vulnerability detection. The performance of DL-based methods usually relies on the quantity and quality of labeled data. However, the current labeled data are generally automatically collected, such as crawled from human-generated commits, making it hard to ensure the quality of the labels. Prior studies have demonstrated that the non-vulnerable code (i.e., negative labels) tends to be unreliable in commonly-used datasets, while vulnerable code (i.e., positive labels) is more determined. Considering the large numbers of unlabeled data in practice, it is necessary and worth exploring to leverage the positive data and large numbers of unlabeled data for more accurate vulnerability detection. In this paper, we focus on the Positive and Unlabeled (PU) learning problem for vulnerability detection and propose a novel model named PILOT, i.e., PositIve and unlabeled Learning mOdel for vulnerability deTection. PILOT only learns from positive and unlabeled data for vulnerability detection. It mainly contains two modules: (1) A distance-aware label selection module, aiming at generating pseudo-labels for selected unlabeled data, which involves the inter-class distance prototype and progressive fine-tuning; (2) A mixed-supervision representation learning module to further alleviate the influence of noise and enhance the discrimination of representations., Comment: This paper is accepted by ASE 2023

Learning per-point semantic features from the hierarchical feature pyramid is essential for point cloud semantic segmentation. However, most previous methods suffered from ambiguous region features or failed to refine per-point features effectively, which leads to information loss and ambiguous semantic identification. To resolve this, we propose Retro-FPN to model the per-point feature prediction as an explicit and retrospective refining process, which goes through all the pyramid layers to extract semantic features explicitly for each point. Its key novelty is a retro-transformer for summarizing semantic contexts from the previous layer and accordingly refining the features in the current stage. In this way, the categorization of each point is conditioned on its local semantic pattern. Specifically, the retro-transformer consists of a local cross-attention block and a semantic gate unit. The cross-attention serves to summarize the semantic pattern retrospectively from the previous layer. And the gate unit carefully incorporates the summarized contexts and refines the current semantic features. Retro-FPN is a pluggable neural network that applies to hierarchical decoders. By integrating Retro-FPN with three representative backbones, including both point-based and voxel-based methods, we show that Retro-FPN can significantly improve performance over state-of-the-art backbones. Comprehensive experiments on widely used benchmarks can justify the effectiveness of our design. The source is available at https://github.com/AllenXiangX/Retro-FPN, Comment: Accepted by ICCV 2023