Integration against, and hence sampling from, high-dimensional probability distributions is of essential importance in many application areas and has been an active research area for decades. One approach that has drawn increasing attention in recent years has been the generation of samples from a target distribution $\mathbb{P}_{\mathrm{tar}}$ using transport maps: if $\mathbb{P}_{\mathrm{tar}} = T_\# \mathbb{P}_{\mathrm{ref}}$ is the pushforward of an easily-sampled probability distribution $\mathbb{P}_{\mathrm{ref}}$ under the transport map $T$, then the application of $T$ to $\mathbb{P}_{\mathrm{ref}}$-distributed samples yields $\mathbb{P}_{\mathrm{tar}}$-distributed samples. This paper proposes the application of transport maps not just to random samples, but also to quasi-Monte Carlo points, higher-order nets, and sparse grids in order for the transformed samples to inherit the original convergence rates that are often better than $N^{-1/2}$, $N$ being the number of samples/quadrature nodes. Our main result is the derivation of an explicit transport map for the case that $\mathbb{P}_{\mathrm{tar}}$ is a mixture of simple distributions, e.g.\ a Gaussian mixture, in which case application of the transport map $T$ requires the solution of an \emph{explicit} ODE with \emph{closed-form} right-hand side. Mixture distributions are of particular applicability and interest since many methods proceed by first approximating $\mathbb{P}_{\mathrm{tar}}$ by a mixture and then sampling from that mixture (often using importance reweighting). Hence, this paper allows for the sampling step to provide a better convergence rate than $N^{-1/2}$ for all such methods.
相關內容
Spatially misaligned data, where the response and covariates are observed at different spatial locations, commonly arise in many environmental studies. Much of the statistical literature on handling spatially misaligned data has been devoted to the case of a single covariate and a linear relationship between the response and this covariate. Motivated by spatially misaligned data collected on air pollution and weather in China, we propose a cokrig-and-regress (CNR) method to estimate spatial regression models involving multiple covariates and potentially non-linear associations. The CNR estimator is constructed by replacing the unobserved covariates (at the response locations) by their cokriging predictor derived from the observed but misaligned covariates under a multivariate Gaussian assumption, where a generalized Kronecker product covariance is used to account for spatial correlations within and between covariates. A parametric bootstrap approach is employed to bias-correct the CNR estimates of the spatial covariance parameters and for uncertainty quantification. Simulation studies demonstrate that CNR outperforms several existing methods for handling spatially misaligned data, such as nearest-neighbor interpolation. Applying CNR to the spatially misaligned air pollution and weather data in China reveals a number of non-linear relationships between PM$_{2.5}$ concentration and several meteorological covariates.
Across various sectors such as healthcare, criminal justice, national security, finance, and technology, large-scale machine learning (ML) and artificial intelligence (AI) systems are being deployed to make critical data-driven decisions. Many have asked if we can and should trust these ML systems to be making these decisions. Two critical components are prerequisites for trust in ML systems: interpretability, or the ability to understand why the ML system makes the decisions it does, and fairness, which ensures that ML systems do not exhibit bias against certain individuals or groups. Both interpretability and fairness are important and have separately received abundant attention in the ML literature, but so far, there have been very few methods developed to directly interpret models with regard to their fairness. In this paper, we focus on arguably the most popular type of ML interpretation: feature importance scores. Inspired by the use of decision trees in knowledge distillation, we propose to leverage trees as interpretable surrogates for complex black-box ML models. Specifically, we develop a novel fair feature importance score for trees that can be used to interpret how each feature contributes to fairness or bias in trees, tree-based ensembles, or tree-based surrogates of any complex ML system. Like the popular mean decrease in impurity for trees, our Fair Feature Importance Score is defined based on the mean decrease (or increase) in group bias. Through simulations as well as real examples on benchmark fairness datasets, we demonstrate that our Fair Feature Importance Score offers valid interpretations for both tree-based ensembles and tree-based surrogates of other ML systems.
With the growing privacy concerns in recommender systems, recommendation unlearning, i.e., forgetting the impact of specific learned targets, is getting increasing attention. Existing studies predominantly use training data, i.e., model inputs, as the unlearning target. However, we find that attackers can extract private information, i.e., gender, race, and age, from a trained model even if it has not been explicitly encountered during training. We name this unseen information as attribute and treat it as the unlearning target. To protect the sensitive attribute of users, Attribute Unlearning (AU) aims to degrade attacking performance and make target attributes indistinguishable. In this paper, we focus on a strict but practical setting of AU, namely Post-Training Attribute Unlearning (PoT-AU), where unlearning can only be performed after the training of the recommendation model is completed. To address the PoT-AU problem in recommender systems, we design a two-component loss function that consists of i) distinguishability loss: making attribute labels indistinguishable from attackers, and ii) regularization loss: preventing drastic changes in the model that result in a negative impact on recommendation performance. Specifically, we investigate two types of distinguishability measurements, i.e., user-to-user and distribution-to-distribution. We use the stochastic gradient descent algorithm to optimize our proposed loss. Extensive experiments on three real-world datasets demonstrate the effectiveness of our proposed methods.
Many researchers and organizations, such as WHO and UNICEF, have raised awareness of the dangers of advertisements targeted at children. While most existing laws only regulate ads on television that may reach children, lawmakers have been working on extending regulations to online advertising and, for example, forbid (e.g., the DSA) or restrict (e.g., the COPPA) advertising based on profiling to children. At first sight, ad platforms such as Google seem to protect children by not allowing advertisers to target their ads to users who are less than 18 years old. However, this paper shows that other targeting features can be exploited to reach children. For example, on YouTube, advertisers can target their ads to users watching a particular video through placement-based targeting, a form of contextual targeting. Hence, advertisers can target children by placing their ads in children-focused videos. Through a series of ad experiments, we show that placement-based targeting is possible on children-focused videos and enables marketing to children. In addition, our ad experiments show that advertisers can use targeting based on profiling (e.g., interest, location, behavior) in combination with placement-based advertising on children-focused videos. We discuss the lawfulness of these two practices concerning DSA and COPPA. Finally, we investigate to which extent real-world advertisers are employing placement-based targeting to reach children with ads on YouTube. We propose a measurement methodology consisting of building a Chrome extension to capture ads and instrument six browser profiles to watch children-focused videos. Our results show that 7% of ads that appear in the children-focused videos we test use placement-based targeting. Hence, targeting children with ads on YouTube is not only hypothetically possible but also occurs in practice...
Fairness or equity in machine learning is profoundly important for societal well-being, but limited public datasets hinder its progress, especially in the area of medicine. It is undeniable that fairness in medicine is one of the most important areas for fairness learning's applications. Currently, no large-scale public medical datasets with 3D imaging data for fairness learning are available, while 3D imaging data in modern clinics are standard tests for disease diagnosis. In addition, existing medical fairness datasets are actually repurposed datasets, and therefore they typically have limited demographic identity attributes with at most three identity attributes of age, gender, and race for fairness modeling. To address this gap, we introduce our Eye Fairness dataset with 30,000 subjects (Harvard-EF) covering three major eye diseases including age-related macular degeneration, diabetic retinopathy, and glaucoma affecting 380 million patients globally. Our Harvard-EF dataset includes both 2D fundus photos and 3D optical coherence tomography scans with six demographic identity attributes including age, gender, race, ethnicity, preferred language, and marital status. We also propose a fair identity scaling (FIS) approach combining group and individual scaling together to improve model fairness. Our FIS approach is compared with various state-of-the-art fairness learning methods with superior performance in the racial, gender, and ethnicity fairness tasks with 2D and 3D imaging data, which demonstrate the utilities of our Harvard-EF dataset for fairness learning. To facilitate fairness comparisons between different models, we propose performance-scaled disparity measures, which can be used to compare model fairness accounting for overall performance levels. The dataset and code are publicly accessible via //ophai.hms.harvard.edu/datasets/harvard-ef30k.
Face recognition technology has advanced significantly in recent years due largely to the availability of large and increasingly complex training datasets for use in deep learning models. These datasets, however, typically comprise images scraped from news sites or social media platforms and, therefore, have limited utility in more advanced security, forensics, and military applications. These applications require lower resolution, longer ranges, and elevated viewpoints. To meet these critical needs, we collected and curated the first and second subsets of a large multi-modal biometric dataset designed for use in the research and development (R&D) of biometric recognition technologies under extremely challenging conditions. Thus far, the dataset includes more than 350,000 still images and over 1,300 hours of video footage of approximately 1,000 subjects. To collect this data, we used Nikon DSLR cameras, a variety of commercial surveillance cameras, specialized long-rage R&D cameras, and Group 1 and Group 2 UAV platforms. The goal is to support the development of algorithms capable of accurately recognizing people at ranges up to 1,000 m and from high angles of elevation. These advances will include improvements to the state of the art in face recognition and will support new research in the area of whole-body recognition using methods based on gait and anthropometry. This paper describes methods used to collect and curate the dataset, and the dataset's characteristics at the current stage.
Deep neural models in recent years have been successful in almost every field, including extremely complex problem statements. However, these models are huge in size, with millions (and even billions) of parameters, thus demanding more heavy computation power and failing to be deployed on edge devices. Besides, the performance boost is highly dependent on redundant labeled data. To achieve faster speeds and to handle the problems caused by the lack of data, knowledge distillation (KD) has been proposed to transfer information learned from one model to another. KD is often characterized by the so-called `Student-Teacher' (S-T) learning framework and has been broadly applied in model compression and knowledge transfer. This paper is about KD and S-T learning, which are being actively studied in recent years. First, we aim to provide explanations of what KD is and how/why it works. Then, we provide a comprehensive survey on the recent progress of KD methods together with S-T frameworks typically for vision tasks. In general, we consider some fundamental questions that have been driving this research area and thoroughly generalize the research progress and technical details. Additionally, we systematically analyze the research status of KD in vision applications. Finally, we discuss the potentials and open challenges of existing methods and prospect the future directions of KD and S-T learning.
Generative commonsense reasoning which aims to empower machines to generate sentences with the capacity of reasoning over a set of concepts is a critical bottleneck for text generation. Even the state-of-the-art pre-trained language generation models struggle at this task and often produce implausible and anomalous sentences. One reason is that they rarely consider incorporating the knowledge graph which can provide rich relational information among the commonsense concepts. To promote the ability of commonsense reasoning for text generation, we propose a novel knowledge graph augmented pre-trained language generation model KG-BART, which encompasses the complex relations of concepts through the knowledge graph and produces more logical and natural sentences as output. Moreover, KG-BART can leverage the graph attention to aggregate the rich concept semantics that enhances the model generalization on unseen concept sets. Experiments on benchmark CommonGen dataset verify the effectiveness of our proposed approach by comparing with several strong pre-trained language generation models, particularly KG-BART outperforms BART by 5.80, 4.60, in terms of BLEU-3, 4. Moreover, we also show that the generated context by our model can work as background scenarios to benefit downstream commonsense QA tasks.
Many tasks in natural language processing can be viewed as multi-label classification problems. However, most of the existing models are trained with the standard cross-entropy loss function and use a fixed prediction policy (e.g., a threshold of 0.5) for all the labels, which completely ignores the complexity and dependencies among different labels. In this paper, we propose a meta-learning method to capture these complex label dependencies. More specifically, our method utilizes a meta-learner to jointly learn the training policies and prediction policies for different labels. The training policies are then used to train the classifier with the cross-entropy loss function, and the prediction policies are further implemented for prediction. Experimental results on fine-grained entity typing and text classification demonstrate that our proposed method can obtain more accurate multi-label classification results.
Object detection typically assumes that training and test data are drawn from an identical distribution, which, however, does not always hold in practice. Such a distribution mismatch will lead to a significant performance drop. In this work, we aim to improve the cross-domain robustness of object detection. We tackle the domain shift on two levels: 1) the image-level shift, such as image style, illumination, etc, and 2) the instance-level shift, such as object appearance, size, etc. We build our approach based on the recent state-of-the-art Faster R-CNN model, and design two domain adaptation components, on image level and instance level, to reduce the domain discrepancy. The two domain adaptation components are based on H-divergence theory, and are implemented by learning a domain classifier in adversarial training manner. The domain classifiers on different levels are further reinforced with a consistency regularization to learn a domain-invariant region proposal network (RPN) in the Faster R-CNN model. We evaluate our newly proposed approach using multiple datasets including Cityscapes, KITTI, SIM10K, etc. The results demonstrate the effectiveness of our proposed approach for robust object detection in various domain shift scenarios.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191