NeurIPS 2024 Papers

In offline reinforcement learning (RL), addressing the out-of-distribution (OOD) action issue has been a focus, but we argue that there exists an OOD state issue that also impairs performance yet has been underexplored. Such an issue describes the scenario when the agent encounters states out of the offline dataset during the test phase, leading to uncontrolled behavior and performance degradation. To this end, we propose SCAS, a simple yet effective approach that unifies OOD state correction and OOD action suppression in offline RL. Technically, SCAS achieves value-aware OOD state correction, capable of correcting the agent from OOD states to high-value in-distribution states. Theoretical and empirical results show that SCAS also exhibits the effect of suppressing OOD actions. On standard offline RL benchmarks, SCAS achieves excellent performance without additional hyperparameter tuning. Moreover, benefiting from its OOD state correction feature, SCAS demonstrates enhanced robustness against environmental perturbations.

The widespread use of black box prediction methods has sparked an increasing interest in algorithm/model-agnostic approaches for quantifying goodness-of-fit, with direct ties to specification testing, model selection and variable importance assessment. A commonly used framework involves defining a predictiveness criterion, applying a cross-fitting procedure to estimate the predictiveness, and utilizing the difference in estimated predictiveness between two models as the test statistic. However, even after standardization, the test statistic typically fails to converge to a non-degenerate distribution under the null hypothesis of equal goodness, leading to what is known as the degeneracy issue. To addresses this degeneracy issue, we present a simple yet effective device, Zipper. It draws inspiration from the strategy of additional splitting of testing data, but encourages an overlap between two testing data splits in predictiveness evaluation. Zipper binds together the two overlapping splits using a slider parameter that controls the proportion of overlap. Our proposed test statistic follows an asymptotically normal distribution under the null hypothesis for any fixed slider value, guaranteeing valid size control while enhancing power by effective data reuse. Finite-sample experiments demonstrate that our procedure, with a simple choice of the slider, works well across a wide range of settings.

Deep learning on tabular data, and particularly tabular representation learning, has recently gained growing interest. However, representation learning for relational databases with multiple tables is still an underexplored area, which may be attributed to the lack of openly available resources. To support the development of foundation models for tabular data and relational databases, we introduce WikiDBs, a novel open-source corpus of 100,000 relational databases. Each database consists of multiple tables connected by foreign keys. The corpus is based on Wikidata and aims to follow certain characteristics of real-world databases. In this paper, we describe the dataset and our method for creating it. By making our code publicly available, we enable others to create tailored versions of the dataset, for example, by creating databases in different languages. Finally, we conduct a set of initial experiments to showcase how WikiDBs can be used to train for data engineering tasks, such as missing value imputation and column type annotation.

Multi-Modal Large Language Models (MLLMs) have demonstrated impressive performance in various VQA tasks. However, they often lack interpretability and struggle with complex visual inputs, especially when the resolution of the input image is high or when the interested region that could provide key information for answering the question is small. To address these challenges, we collect and introduce the large-scale Visual CoT dataset comprising 438k question-answer pairs, annotated with intermediate bounding boxes highlighting key regions essential for answering the questions. Additionally, about 98k pairs of them are annotated with detailed reasoning steps. Importantly, we propose a multi-turn processing pipeline that dynamically focuses on visual inputs and provides interpretable thoughts. We also introduce the related benchmark to evaluate the MLLMs in scenarios requiring specific local region identification. Extensive experiments demonstrate the effectiveness of our framework and shed light on better inference strategies. The Visual CoT dataset, benchmark, and pre-trained models are available on this [website](https://hao-shao.com/projects/viscot.html) to support further research in this area.

We present a new random walk for uniformly sampling high-dimensional convex bodies. It achieves state-of-the-art runtime complexity with stronger guarantees on the output than previously known, namely in Rényi divergence (which implies TV, $\mathcal{W}_2$, KL, $\chi^2$). The proof departs from known approaches for polytime algorithms for the problem - we utilize a stochastic diffusion perspective to show contraction to the target distribution with the rate of convergence determined by functional isoperimetric constants of the stationary density.

The need for large text corpora has increased with the advent of pretrained language models and, in particular, the discovery of scaling laws for these models. Most available corpora have sufficient data only for languages with large dominant communities. However, there is no corpus available that (i) covers a wide range of minority languages; (ii) is generated by an open-source reproducible pipeline; and (iii) is rigorously cleaned from noise, making it trustworthy to use. We present GlotCC, a clean, document-level, 2TB general domain corpus derived from CommonCrawl, covering more than 1000 languages. We make GlotCC and the system used to generate it— including the pipeline, language identification model, and filters—available to the research community. Corpus v. 1.0 https://huggingface.co/datasets/cis-lmu/GlotCC-v1 Pipeline v. 3.0 https://github.com/cisnlp/GlotCC

Correctly capturing the symmetry transformations of data can lead to efficient models with strong generalization capabilities, though methods incorporating symmetries often require prior knowledge. While recent advancements have been made in learning those symmetries directly from the dataset, most of this work has focused on the discriminative setting. In this paper, we take inspiration from group theoretic ideas to construct a generative model that explicitly aims to capture the data's approximate symmetries. This results in a model that, given a prespecified broad set of possible symmetries, learns to what extent, if at all, those symmetries are actually present. Our model can be seen as a generative process for data augmentation. We provide a simple algorithm for learning our generative model and empirically demonstrate its ability to capture symmetries under affine and color transformations, in an interpretable way. Combining our symmetry model with standard generative models results in higher marginal test-log-likelihoods and improved data efficiency.

Fourier Neural Operators (FNOs) have shown promise for solving partial differential equations (PDEs). Typically, FNOs employ separate parameters for different frequency modes to specify tunable kernel integrals in Fourier space, which, yet, results in an undesirably large number of parameters when solving high-dimensional PDEs. A workaround is to abandon the frequency modes exceeding a predefined threshold, but this limits the FNOs' ability to represent high-frequency details and poses non-trivial challenges for hyper-parameter specification. To address these, we propose AMortized Fourier Neural Operator (AM-FNO), where an amortized neural parameterization of the kernel function is deployed to accommodate arbitrarily many frequency modes using a fixed number of parameters. We introduce two implementations of AM-FNO, based on the recently developed, appealing Kolmogorov–Arnold Network (KAN) and Multi-Layer Perceptrons (MLPs) equipped with orthogonal embedding functions respectively. We extensively evaluate our method on diverse datasets from various domains and observe up to 31\% average improvement compared to competing neural operator baselines.

To build photorealistic avatars that users can embody, human modelling must be complete (cover the full body), driveable (able to reproduce the current motion and appearance from the user), and generalizable (_i.e._, easily adaptable to novel identities). Towards these goals, _paired_ captures, that is, captures of the same subject obtained from systems of diverse quality and availability, are crucial. However, paired captures are rarely available to researchers outside of dedicated industrial labs: _Codec Avatar Studio_ is our proposal to close this gap. Towards generalization and driveability, we introduce a dataset of 256 subjects captured in two modalities: high resolution multi-view scans of their heads, and video from the internal cameras of a headset. Towards completeness, we introduce a dataset of 4 subjects captured in eight modalities: high quality relightable multi-view captures of heads and hands, full body multi-view captures with minimal and regular clothes, and corresponding head, hands and body phone captures. Together with our data, we also provide code and pre-trained models for different state-of-the-art human generation models. Our datasets and code are available at https://github.com/facebookresearch/ava-256 and https://github.com/facebookresearch/goliath.

Data is a critical resource for machine learning (ML), yet working with data remains a key friction point. This paper introduces Croissant, a metadata format for datasets that creates a shared representation across ML tools, frameworks, and platforms. Croissant makes datasets more discoverable, portable, and interoperable, thereby addressing significant challenges in ML data management. Croissant is already supported by several popular dataset repositories, spanning hundreds of thousands of datasets, enabling easy loading into the most commonly-used ML frameworks, regardless of where the data is stored. Our initial evaluation by human raters shows that Croissant metadata is readable, understandable, complete, yet concise.

Critique ability, i.e., the capability of Large Language Models (LLMs) to identify and rectify flaws in responses, is crucial for their applications in self-improvement and scalable oversight. While numerous studies have been proposed to evaluate critique ability of LLMs, their comprehensiveness and reliability are still limited. To overcome this problem, we introduce CriticEval, a novel benchmark designed to comprehensively and reliably evaluate critique ability of LLMs. Specifically, to ensure the comprehensiveness, CriticEval evaluates critique ability from four dimensions across nine diverse task scenarios. It evaluates both scalar-valued and textual critiques, targeting responses of varying quality. To ensure the reliability, a large number of critiques are annotated to serve as references, enabling GPT-4 to evaluate textual critiques reliably. Extensive evaluations of open-source and closed-source LLMs first validate the reliability of evaluation in CriticEval. Then, experimental results demonstrate the promising potential of open-source LLMs, the effectiveness of critique datasets and several intriguing relationships between the critique ability and some critical factors, including task types, response qualities and critique dimensions.

Many tasks in explainable machine learning, such as data valuation and feature attribution, perform expensive computation for each data point and are intractable for large datasets. These methods require efficient approximations, and although amortizing the process by learning a network to directly predict the desired output is a promising solution, training such models with exact labels is often infeasible. We therefore explore training amortized models with noisy labels, and we find that this is inexpensive and surprisingly effective. Through theoretical analysis of the label noise and experiments with various models and datasets, we show that this approach tolerates high noise levels and significantly accelerates several feature attribution and data valuation methods, often yielding an order of magnitude speedup over existing approaches.

Normalization layers have recently experienced a renaissance in the deep reinforcement learning and continual learning literature, with several works highlighting diverse benefits such as improving loss landscape conditioning and combatting overestimation bias. However, normalization brings with it a subtle but important side effect: an equivalence between growth in the norm of the network parameters and decay in the effective learning rate. This becomes problematic in continual learning settings, where the resulting learning rate schedule may decay to near zero too quickly relative to the timescale of the learning problem. We propose to make the learning rate schedule explicit with a simple re-parameterization which we call Normalize-and-Project (NaP), which couples the insertion of normalization layers with weight projection, ensuring that the effective learning rate remains constant throughout training. This technique reveals itself as a powerful analytical tool to better understand learning rate schedules in deep reinforcement learning, and as a means of improving robustness to nonstationarity in synthetic plasticity loss benchmarks along with both the single-task and sequential variants of the Arcade Learning Environment. We also show that our approach can be easily applied to popular architectures such as ResNets and transformers while recovering and in some cases even slightly improving the performance of the base model in common stationary benchmarks.

Datasets play a pivotal role in training visual models, facilitating the development of abstract understandings of visual features through diverse image samples and multidimensional attributes. However, in the realm of aesthetic evaluation of artistic images, datasets remain relatively scarce. Existing painting datasets are often characterized by limited scoring dimensions and insufficient annotations, thereby constraining the advancement and application of automatic aesthetic evaluation methods in the domain of painting. To bridge this gap, we introduce the Aesthetics Paintings and Drawings Dataset (APDD), the first comprehensive collection of paintings encompassing 24 distinct artistic categories and 10 aesthetic attributes. Building upon the initial release of APDDv1, our ongoing research has identified opportunities for enhancement in data scale and annotation precision. Consequently, APDDv2 boasts an expanded image corpus and improved annotation quality, featuring detailed language comments to better cater to the needs of both researchers and practitioners seeking high-quality painting datasets. Furthermore, we present an updated version of the Art Assessment Network for Specific Painting Styles, denoted as ArtCLIP. Experimental validation demonstrates the superior performance of this revised model in the realm of aesthetic evaluation, surpassing its predecessor in accuracy and efficacy. The dataset and model are available at https://github.com/BestiVictory/APDDv2.git.

Learning generalized models from biased data is an important undertaking toward fairness in deep learning. To address this issue, recent studies attempt to identify and leverage bias-conflicting samples free from spurious correlations without prior knowledge of bias or an unbiased set. However, spurious correlation remains an ongoing challenge, primarily due to the difficulty in correctly detecting these samples. In this paper, inspired by the similarities between mislabeled samples and bias-conflicting samples, we approach this challenge from a novel perspective of mislabeled sample detection. Specifically, we delve into Influence Function, one of the standard methods for mislabeled sample detection, for identifying bias-conflicting samples and propose a simple yet effective remedy for biased models by leveraging them. Through comprehensive analysis and experiments on diverse datasets, we demonstrate that our new perspective can boost the precision of detection and rectify biased models effectively. Furthermore, our approach is complementary to existing methods, showing performance improvement even when applied to models that have already undergone recent debiasing techniques.

Multi-person pose estimation and tracking have been actively researched by the computer vision community due to their practical applicability. However, existing human pose estimation and tracking datasets have only been successful in typical scenarios, such as those without motion blur or with well-lit conditions. These RGB-based datasets are limited to learning under extreme motion blur situations or poor lighting conditions, making them inherently vulnerable to such scenarios. As a promising solution, bio-inspired event cameras exhibit robustness in extreme scenarios due to their high dynamic range and micro-second level temporal resolution. Therefore, in this paper, we introduce a new hybrid dataset encompassing both RGB and event data for human pose estimation and tracking in two extreme scenarios: low-light and motion blur environments. The proposed Event-guided Human Pose Estimation and Tracking in eXtreme Conditions (EHPT-XC) dataset covers cases of motion blur caused by dynamic objects and low-light conditions individually as well as both simultaneously. With EHPT-XC, we aim to inspire researchers to tackle pose estimation and tracking in extreme conditions by leveraging the advantageous of the event camera. Project pages are available at https://github.com/Chohoonhee/EHPT-XC.

In machine learning research, it is common to evaluate algorithms via their performance on standard benchmark datasets. While a growing body of work establishes guidelines for---and levies criticisms at---data and benchmarking practices in machine learning, comparatively less attention has been paid to the data repositories where these datasets are stored, documented, and shared. In this paper, we analyze the landscape of these _benchmark data repositories_ and the role they can play in improving benchmarking. This role includes addressing issues with both datasets themselves (e.g., representational harms, construct validity) and the manner in which evaluation is carried out using such datasets (e.g., overemphasis on a few datasets and metrics, lack of reproducibility). To this end, we identify and discuss a set of considerations surrounding the design and use of benchmark data repositories, with a focus on improving benchmarking practices in machine learning.

Performance on popular ML benchmarks is highly correlated with model scale, suggesting that most benchmarks tend to measure a similar underlying factor of general model capabilities. However, substantial research effort remains devoted to designing new benchmarks, many of which claim to measure novel phenomena. In the spirit of the Bitter Lesson, we leverage spectral analysis to measure an underlying capabilities component, the direction in benchmark-performance-space which explains most variation in model performance. In an extensive analysis of existing safety benchmarks, we find that variance in model performance on many safety benchmarks is largely explained by the capabilities component. In response, we argue that safety research should prioritize metrics which are not highly correlated with scale. Our work provides a lens to analyze both novel safety benchmarks and novel safety methods, which we hope will enable future work to make differential progress on safety.

Following step-by-step procedures is an essential component of various activities carried out by individuals in their daily lives. These procedures serve as a guiding framework that helps to achieve goals efficiently, whether it is assembling furniture or preparing a recipe. However, the complexity and duration of procedural activities inherently increase the likelihood of making errors. Understanding such procedural activities from a sequence of frames is a challenging task that demands an accurate interpretation of visual information and the ability to reason about the structure of the activity. To this end, we collect a new egocentric 4D dataset, CaptainCook4D, comprising 384 recordings (94.5 hours) of people performing recipes in real kitchen environments. This dataset consists of two distinct types of activity: one in which participants adhere to the provided recipe instructions and another in which they deviate and induce errors. We provide 5.3K step annotations and 10K fine-grained action annotations and benchmark the dataset for the following tasks: error recognition, multistep localization and procedure learning.