Publications

2026

1. NeurIPS

   Is Backpropagation Optimal? When Synthetic Gradients Improve Sample Efficiency

   Yibo Jacky Zhang, Zeyu Tang, and Sanmi Koyejo

   In The Fortieth Annual Conference on Neural Information Processing Systems, 2026.

   ABSTRACT arXiv PDF

   Backpropagation is the default learning rule for artificial neural networks and is often treated as the settled approach whenever differentiability is available. In this work, we revisit this convention through a theoretical lens of sample efficiency. We introduce a unified vectorized feedback framework for loss-based and reward-based learning on computational graphs, in which synthetic gradients emerge as a natural alternative to backpropagation. We characterize the conditions under which synthetic gradients can achieve a lower gradient-estimation mean squared error than backpropagation. We construct examples illustrating that this sample efficiency advantage can be arbitrarily large. Experiments on contextual bandits and reinforcement learning tasks demonstrate the potential of our theoretical findings.

2. NeurIPSE&D Track

   χ-Bench: Can AI Agents Automate End-to-End, Long-Horizon, Policy-Rich Healthcare Workflows?

   Haolin Chen, Deon Metelski, Leon Qi, Tao Xia, Joonyul Lee, Steve Brown, Kevin Riley, Frank Wang, T. Y. Alvin Liu, Hank Capps, Zeyu Tang, Xiangchen Song, Lingjing Kong, Fan Feng, Tianyi Zeng, Zhiwei Liu, Zixian Ma, Hang Jiang, Fangli Geng, Yuan Yuan, Chenyu You, Qingsong Wen, Hua Wei, Yanjie Fu, Yue Zhao, Carl Yang, Biwei Huang, Kun Zhang, Caiming Xiong, Sanmi Koyejo, Eric P. Xing, Philip S. Yu, and Weiran Yao

   In The Fortieth Annual Conference on Neural Information Processing Systems, Evaluations and Datasets Track, 2026.

   ABSTRACT arXiv PDF

   End-to-end automation of realistic healthcare operations stresses three capabilities underrepresented in current benchmarks: policy density, decisions must be grounded in a large library of medical, insurance, and operational rules; multi-role composition, a single task requires the agent to play multiple roles with handoffs; and multilateral interaction: intermediate workflow steps are multi-turn dialogs, such as peer-to-peer review and patient outreach. We introduce χ-Bench, a benchmark of long-horizon healthcare workflows across three domains: provider prior authorization, payer utilization management, and care management. Each task hands the agent a clinical case in a high-fidelity simulator of 20 healthcare apps exposed via 87 MCP tools, which it must drive to a terminal status through tool calls and writing the role’s artifacts, guided by a 1,279-document managed-care operations handbook skill. Across 30 agent harness/model configurations, the best agent resolves only 28.0% of tasks, no agent clears 20% on strict pass^3, and executing all tasks in a single session slumps the performance to 3.8%. These results raise the hypothesis that similar gaps are likely to surface in other policy-dense, role-composed, irreversible enterprise domains.

3. COLM

   In-Situ Behavioral Evaluation for LLM Fairness, Not Standardized-Test Scores

   Zeyu Tang*, Sang T. Truong*, Deonna Owens*, Shreyas Sharma, Yibo Jacky Zhang, Brando Miranda, and Sanmi Koyejo

   In The Third Conference on Language Modeling, 2026.

   ABSTRACT arXiv PDF

   LLM fairness should be evaluated through in-situ behavioral pattern rather than standardized-test Q&A benchmarks. We show that the standardized-test paradigm can be structurally unreliable: surface-level prompt construction choices, although entirely orthogonal to the fairness question being tested, account for the majority of score variance, shift fairness conclusions in both the direction and the magnitude, and result in severe discordance in model rankings. We develop MAC-Fairness, a framework that embeds controlled variation factors into in-situ behavioral evaluation, examining how models’ disparate-treatment behaviors shift when identity is varied as part of natural multi-agent conversation. Repurposing standardized-test questions as conversation seeds rather than as the evaluation instrument, we evaluate within-model differences in position persistence (how they hold positions, from the self-perspective) and peer receptiveness (how receptive they are to peers, from the other-perspective) across 8 million conversation transcripts spanning multiple models and identity presence configurations. In-situ behavioral evaluation reveals stable, model-specific, disparate-treatment behavioral signatures that could generalize across different fairness benchmarks, a form of evidence the standardized-test paradigm does not offer.

4. ICMLPosition

   Position: Beyond Sensitive Attributes, ML Fairness Should Quantify Structural Injustice via Social Determinants

   Zeyu Tang, Alex John London, Atoosa Kasirzadeh, Sarah Stewart de Ramirez, Peter Spirtes, Kun Zhang, and Sanmi Koyejo

   In The Forty-Third International Conference on Machine Learning, 2026.

   ABSTRACT arXiv PDF

   Algorithmic fairness research has largely framed unfairness as discrimination along sensitive attributes. However, this approach limits visibility into unfairness as structural injustice instantiated through social determinants, which are contextual variables that shape attributes and outcomes without pertaining to specific individuals. This position paper argues that the field should quantify structural injustice via social determinants, beyond sensitive attributes. Drawing on cross-disciplinary insights, we argue that prevailing technical paradigms fail to adequately capture unfairness as structural injustice, because contexts are potentially treated as noise to be normalized rather than signal to be audited. We further demonstrate the practical urgency of this shift through a theoretical model of college admissions, a demographic study using U.S. census data, and a high-stakes domain application regarding breast cancer screening within an integrated U.S. healthcare system. Our results indicate that mitigation strategies centered solely on sensitive attributes can introduce new forms of structural injustice. We contend that auditing structural injustice through social determinants must precede mitigation, and call for new technical developments that move beyond sensitive-attribute-centered notions of fairness as non-discrimination.

5. ICMLSpotlight

   Causal Modeling of Selection in Evolution

   Haoyue Dai, Zeyu Tang, Peter Spirtes, and Kun Zhang

   In The Forty-Third International Conference on Machine Learning, 2026.

   ABSTRACT arXiv PDF

   Understanding potential selection in data is crucial for causal discovery; we argue that "selection" in common narratives takes two forms, which we term static and evolutionary selection, respectively. Static selection refers to a one-shot filtering process where observed data consist of a subset of the population of interest, as in survey volunteer bias. Evolutionary selection, in contrast, operates through repeated rounds of differential fitness in reproduction, where observed data constitute the latest generation shaped by a historical trajectory, as in immune adaptation, antibiotic resistance, and social norm emergence. Existing methods largely conflate these two forms and rely on an identical graphical model of selection. We show that this model is valid for static settings but fails to characterize data under evolution, yielding false discovery results. To address this, we introduce a new model that specifically characterizes evolutionary selection, and develop a sound and complete procedure for identifying such models from data across one or multiple environments or generations. Experimental results validate the method’s ability to uncover the relevant mechanisms underlying evolution from data.

6. ACLOral

   Mechanistic Interpretability Should Prioritize Feature Consistency in Sparse Autoencoders

   Xiangchen Song*, Aashiq Muhamed*, Yujia Zheng, Lingjing Kong, Zeyu Tang, Mona T. Diab, Virginia Smith, and Kun Zhang

   In The Sixty-Fourth Annual Meeting of the Association for Computational Linguistics, 2026.

   ABSTRACT

   Sparse Autoencoders (SAEs) are a prominent tool in mechanistic interpretability (MI) for decomposing neural network activations into interpretable features. However, the aspiration to identify a canonical set of features is challenged by the observed inconsistency of learned SAE features across different training runs, undermining reproducibility and complicating model comparison. We study run-to-run feature consistency in SAEs and argue that it should be reported as a standard evaluation axis alongside reconstruction and sparsity. We propose the Pairwise Dictionary Mean Correlation Coefficient (PW-MCC) as an assignment-based metric to quantify consistency and demonstrate that high levels are achievable (PW-MCC ≈ 0.80 for TopK SAEs on LLM activations) with appropriate architectural choices. Our contributions include: (i) theoretical grounding for strong consistency in the idealized setting of TopK SAEs; (ii) synthetic validation using a model organism, which verifies PW-MCC as a reliable proxy for ground-truth recovery; and (iii) empirical analysis on LLM activations, where PW-MCC correlates with the similarity of automatically generated natural-language feature explanations.

7. ICLR

   Selection, Reflection and Self-Refinement: Revisit Reasoning Tasks via a Causal Lens

   Yunlong Deng, Boyang Sun, Yan Li, Lingjing Kong, Zeyu Tang, Kun Zhang, and Guangyi Chen

   In The Fourteenth International Conference on Learning Representations, 2026.

   ABSTRACT arXiv PDF

   Due to their inherent complexity, reasoning tasks have long been regarded as rigorous benchmarks for assessing the capabilities of machine learning models, especially large language models (LLMs). Although humans can solve these tasks with ease, existing models, even after extensive pre-training and post-training at scale, still fail to perform reasoning reliably. In this paper, we revisit reasoning tasks from a causal perspective, seeking to understand their behavior in latent space and to offer insights for addressing their challenges. Specifically, we cast reasoning tasks as a selection mechanism, in which high-level logical concepts function as selection operators on the given observations, such as, identifying the correct answer in a math problem or filling the appropriate entry in Sudoku. We emphasize two key properties of this formulation that shed light on the difficulty of reasoning tasks. First, the latent space exceeds the observation space in complexity, even when the correct answer is fully determined by the observed input. Second, the latent variables, corresponding to logical thought, are densely structured and exhibit strong dependencies. Building on this formulation, we introduce a framework, called SR2, that incorporates the estimated latent variables as feedback into the selection mechanism, thereby facilitating the learning of dense dependencies among latent representations. The framework consists of three key modules: reflective representation learning, dependency self-refinement, and periodic intermediate alignment. Experimentally, we show that our approach yields significant gains in reasoning accuracy, for example, attaining over 10% improvement in performance with 8X fewer parameters on the Sudoku and Maze tasks over the recent advances.

8. ICLR

   Ada-Diffuser: Latent-Aware Adaptive Diffusion for Decision-Making

   Fan Feng, Selena Ge, Minghao Fu, Zijian Li, Yujia Zheng, Zeyu Tang, Yingyao Hu, Biwei Huang, and Kun Zhang

   In The Fourteenth International Conference on Learning Representations, 2026.

   ABSTRACT

   Recent work has framed decision-making as a sequence modeling problem using generative models such as diffusion models. Although promising, these approaches often overlook latent factors that exhibit evolving dynamics, elements that are fundamental to environment transitions, reward structures, and high-level agent behavior. Explicitly modeling these hidden processes is essential for both precise dynamics modeling and effective decision-making. In this paper, we propose a unified framework that explicitly incorporates latent dynamic inference into generative decision-making from minimal yet sufficient observations. We theoretically show that under mild conditions, the latent process can be identified from small temporal blocks of observations. Building on this insight, we introduce Ada-Diffuser, a causal diffusion model that learns the temporal structure of observed interactions and the underlying latent dynamics simultaneously, and furthermore, leverages them for planning and control. With a proper modular design, Ada-Diffuser supports both planning and policy learning tasks, enabling adaptation to latent variations in dynamics, rewards, and even recovering hidden action variables from action-free demonstrations. Extensive experiments on locomotion and robotic manipulation benchmarks demonstrate the model’s effectiveness in accurate latent inference, long-horizon planning, and adaptive policy learning.

9. ICLR

   Controllable Video Generation with Provable Disentanglement

   Yifan Shen*, Peiyuan Zhu*, Zijian Li, Shaoan Xie, Namrata Deka, Zongfang Liu, Zeyu Tang, Guangyi Chen, and Kun Zhang

   In The Fourteenth International Conference on Learning Representations, 2026.

   ABSTRACT arXiv PDF

   Recent work has framed decision-making as a sequence modeling problem using generative models such as diffusion models. Although promising, these approaches often overlook latent factors that exhibit evolving dynamics, elements that are fundamental to environment transitions, reward structures, and high-level agent behavior. Explicitly modeling these hidden processes is essential for both precise dynamics modeling and effective decision-making. In this paper, we propose a unified framework that explicitly incorporates latent dynamic inference into generative decision-making from minimal yet sufficient observations. We theoretically show that under mild conditions, the latent process can be identified from small temporal blocks of observations. Building on this insight, we introduce Ada-Diffuser, a causal diffusion model that learns the temporal structure of observed interactions and the underlying latent dynamics simultaneously, and furthermore, leverages them for planning and control. With a proper modular design, Ada-Diffuser supports both planning and policy learning tasks, enabling adaptation to latent variations in dynamics, rewards, and even recovering hidden action variables from action-free demonstrations. Extensive experiments on locomotion and robotic manipulation benchmarks demonstrate the model’s effectiveness in accurate latent inference, long-horizon planning, and adaptive policy learning.

2025

1. NeurIPS

   Towards Self-Refinement of Vision-Language Models with Triangular Consistency

   Yunlong Deng*, Guangyi Chen*, Tianpei Gu, Lingjing Kong, Yan Li, Zeyu Tang, and Kun Zhang

   In The Thirty-Ninth Annual Conference on Neural Information Processing Systems, 2025.

   ABSTRACT arXiv PDF

   Vision-Language Models (VLMs) integrate visual knowledge with the analytical capabilities of Large Language Models (LLMs) through supervised visual instruction tuning, using image-question-answer triplets. However, the potential of VLMs trained without supervised instruction remains largely unexplored. This study validates that VLMs possess inherent self-refinement capabilities, enabling them to generate high-quality supervised data without external inputs and thereby learn autonomously. Specifically, to stimulate the self-refinement ability of VLMs, we propose a self-refinement framework based on a Triangular Consistency principle: within the image-query-answer triangle, any masked elements should be consistently and accurately reconstructed. The framework involves three steps: (1) We enable the instruction generation ability of VLMs by adding multi-task instruction tuning like image -> question-answer or image-answer -> question. (2) We generate image-query-answer triplets from unlabeled images and use the Triangular Consistency principle for filtering. (3) The model is further updated using the filtered synthetic data. To investigate the underlying mechanisms behind this self-refinement capability, we conduct a theoretical analysis from a causal perspective. Using the widely recognized LLaVA-1.5 as our baseline, our experiments reveal that the model can autonomously achieve consistent, though deliberately modest, improvements across multiple benchmarks without any external supervision, such as human annotations or environmental feedback. We expect that the insights of this study on the self-refinement ability of VLMs can inspire future research on the learning mechanism of VLMs.

2. NeurIPSD&B Track

   Fantastic Bugs and Where to Find Them in AI Benchmarks

   Sang T. Truong, Yuheng Tu, Michael Hardy, Anka Reuel, Zeyu Tang, Jirayu Burapacheep, Jonathan Perera, Chibuike Uwakwe, Ben Domingue, Nick Haber, and Sanmi Koyejo

   In The Thirty-Ninth Annual Conference on Neural Information Processing Systems, Datasets and Benchmarks Track, 2025.

   ABSTRACT arXiv PDF

   Benchmarks are pivotal in driving AI progress, and invalid benchmark questions frequently undermine their reliability. Manually identifying and correcting errors among thousands of benchmark questions is not only infeasible but also a critical bottleneck for reliable evaluation. In this work, we introduce a framework for systematic benchmark revision that leverages statistical analysis of response patterns to flag potentially invalid questions for further expert review. Our approach builds on a core assumption commonly used in AI evaluations that the mean score sufficiently summarizes model performance. This implies a unidimensional latent construct underlying the measurement experiment, yielding expected ranges for various statistics for each item. When empirically estimated values for these statistics fall outside the expected range for an item, the item is more likely to be problematic. Across nine widely used benchmarks, our method guides expert review to identify problematic questions with up to 84% precision. In addition, we introduce an LLM-judge first pass to review questions, further reducing human effort. Together, these components provide an efficient and scalable framework for systematic benchmark revision.

3. ICML

   Reflection-Window Decoding: Text Generation with Selective Refinement

   Zeyu Tang*, Zhenhao Chen*, Xiangchen Song, Loka Li, Yunlong Deng, Yifan Shen, Guangyi Chen, Peter Spirtes, and Kun Zhang

   In The Forty-Second International Conference on Machine Learning, 2025.

   ABSTRACT arXiv PDF

   The autoregressive decoding for text generation in large language models (LLMs), while widely used, is inherently suboptimal due to the lack of a built-in mechanism to perform refinement and/or correction of the generated content. In this paper, we consider optimality in terms of the joint probability over the generated response, when jointly considering all tokens at the same time. We theoretically characterize the potential deviation of the autoregressively generated response from its globally optimal counterpart that is of the same length. Our analysis suggests that we need to be cautious when noticeable uncertainty arises during text generation, which may signal the sub-optimality of the generation history. To address the pitfall of autoregressive decoding for text generation, we propose an approach that incorporates a sliding reflection window and a pausing criterion, such that refinement and generation can be carried out interchangeably as the decoding proceeds. Our selective refinement framework strikes a balance between efficiency and optimality, and our extensive experimental results demonstrate the effectiveness of our approach.

4. ICML

   Fairness on Principal Stratum: A New Perspective on Counterfactual Fairness

   Haoxuan Li, Zeyu Tang, Zhichao Jiang, Zhuangyan Fang, Yue Liu, Zhi Geng, and Kun Zhang

   In The Forty-Second International Conference on Machine Learning, 2025.

   ABSTRACT PDF

   Fairness in human and algorithmic decision-making is crucial in areas such as criminal justice, education, and social welfare. Recently, counterfactual fairness has drawn increasing research interest, suggesting that decision-making for individuals should remain the same when intervening with different values on protected attributes. Nevertheless, the question of "which attributes and individuals should be protected" is rarely discussed in the existing counterfactual fairness literature. For example, when considering leg disability as a protected attribute, the algorithms should not treat individuals with leg disabilities differently in college admissions, but one may naturally consider this factor when selecting runner athletes. In other words, when and how to enforce fairness is expected to depend on the causal relation between the protected attribute and the outcome of interest. Formally, this paper proposes principal counterfactual fairness using the concept of principal stratification from the causal inference literature, focusing on whether an algorithm is counterfactually fair for individuals whose protected attribute has no individual causal effect on the outcome of interest. To examine whether an algorithm satisfies principal counterfactual fairness, we derive the statistical bounds and propose a post-processing approach to achieving principal counterfactual fairness with minimal individual decision changes. Experiments are conducted using synthetic and real-world datasets to verify the effectiveness of our methods.

5. ICML

   Learning Vision and Language Concepts for Controllable Image Generation

   Shaoan Xie*, Lingjing Kong*, Yujia Zheng, Zeyu Tang, Eric P. Xing, Guangyi Chen, and Kun Zhang

   In The Forty-Second International Conference on Machine Learning, 2025.

   ABSTRACT PDF

   Concept learning seeks to extract semantic and interpretable representations of atomic concepts from high-dimensional data such as images and text, which can be instrumental to a variety of downstream tasks (e.g., image generation/editing). Despite its importance, the theoretical foundations for learning atomic concepts and their interactions, especially from multimodal distributions, remain underexplored. In this work, we establish fundamental conditions on learning atomic multimodal concepts and their underlying interactions. We formulate concept learning as a latent variable identification problem, representing atomic concepts and their interactions as latent variables within a graphical model. Our theoretical contribution is providing component-wise identifiability of atomic concepts under flexible, nonparametric conditions that accommodate both continuous and discrete modalities. This advances beyond prior approaches that offer only block-wise identifiability or rely on parametric assumptions.

6. TMLR

   Long-Term Fairness Inquiries and Pursuits in Machine Learning: A Survey of Notions, Methods, and Challenges

   Usman Gohar, Zeyu Tang, Jialu Wang, Kun Zhang, Peter Spirtes, Yang Liu, and Lu Cheng

   Transactions on Machine Learning Research, 2025.

   ABSTRACT arXiv PDF

   The widespread integration of Machine Learning systems in daily life, particularly in high-stakes domains, has raised concerns about the fairness implications. While prior works have investigated static fairness measures, recent studies reveal that automated decision-making has long-term implications and that off-the-shelf fairness approaches may not serve the purpose of achieving long-term fairness. Additionally, the existence of feedback loops and the interaction between models and the environment introduces additional complexities that may deviate from the initial fairness goals. In this survey, we review existing literature on long-term fairness from different perspectives and present a taxonomy for long-term fairness studies. We highlight key challenges and consider future research directions, analyzing both current issues and potential further explorations.

7. ICLR

   Prompting Fairness: Integrating Causality to Debias Large Language Models

   Jingling Li*, Zeyu Tang*, Xiaoyu Liu, Peter Spirtes, Kun Zhang, Liu Leqi, and Yang Liu

   In The Thirteenth International Conference on Learning Representations (preliminary version titled "Steering LLMs Towards Unbiased Responses: A Causality-Guided Debiasing Framework"), 2025.

   ABSTRACT arXiv PDF

   Large language models (LLMs), despite their remarkable capabilities, are susceptible to generating biased and discriminatory responses. As LLMs increasingly influence high-stakes decision-making (e.g., hiring and healthcare), mitigating these biases becomes critical. In this work, we propose a causality-guided debiasing framework to tackle social biases, aiming to reduce the objectionable dependence between LLMs’ decisions and the social information in the input. Our framework introduces a novel perspective to identify how social information can affect an LLM’s decision through different causal pathways. Leveraging these causal insights, we outline principled prompting strategies that regulate these pathways through selection mechanisms. This framework not only unifies existing prompting-based debiasing techniques, but also opens up new directions for reducing bias by encouraging the model to prioritize fact-based reasoning over reliance on biased social cues. We validate our framework through extensive experiments on real-world datasets across multiple domains, demonstrating its effectiveness in debiasing LLM decisions, even with only black-box access to the model.

8. ICLROral

   When Selection Meets Intervention: Additional Complexities in Causal Discovery

   Haoyue Dai, Ignavier Ng, Jianle Sun, Zeyu Tang, Gongxu Luo, Xinshuai Dong, Peter Spirtes, and Kun Zhang

   In The Thirteenth International Conference on Learning Representations, 2025.

   ABSTRACT arXiv PDF

   We address the common yet often-overlooked selection bias in interventional studies, where subjects are selectively enrolled into experiments. For instance, participants in a drug trial are usually patients of the relevant disease; A/B tests on mobile applications target existing users only, and gene perturbation studies typically focus on specific cell types, such as cancer cells. Ignoring this bias leads to incorrect causal discovery results. Even when recognized, the existing paradigm for interventional causal discovery still fails to address it. This is because subtle differences in when and where interventions happen can lead to significantly different statistical patterns. We capture this dynamic by introducing a graphical model that explicitly accounts for both the observed world (where interventions are applied) and the counterfactual world (where selection occurs while interventions have not been applied). We characterize the Markov property of the model, and propose a provably sound algorithm to identify causal relations as well as selection mechanisms up to the equivalence class, from data with soft interventions and unknown targets. Through synthetic and real-world experiments, we demonstrate that our algorithm effectively identifies true causal relations despite the presence of selection bias.

9. CVPRHighlight

   SmartCLIP: Modular Vision-language Alignment with Identification Guarantees

   Shaoan Xie*, Lingjing Kong*, Yujia Zheng, Yu Yao, Zeyu Tang, Eric P. Xing, Guangyi Chen, and Kun Zhang

   In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2025.

   ABSTRACT arXiv PDF

   Contrastive Language-Image Pre-training (CLIP) has emerged as a pivotal model in computer vision and multimodal learning, achieving state-of-the-art performance at aligning visual and textual representations through contrastive learning. However, CLIP struggles with potential information misalignment in many image-text datasets and suffers from entangled representation. On the one hand, short captions for a single image in datasets like MSCOCO may describe disjoint regions in the image, leaving the model uncertain about which visual features to retain or disregard. On the other hand, directly aligning long captions with images can lead to the retention of entangled details, preventing the model from learning disentangled, atomic concepts – ultimately limiting its generalization on certain downstream tasks involving short prompts. In this paper, we establish theoretical conditions that enable flexible alignment between textual and visual representations across varying levels of granularity. Specifically, our framework ensures that a model can not only \emphpreserve cross-modal semantic information in its entirety but also \emphdisentangle visual representations to capture fine-grained textual concepts. Building on this foundation, we introduce \ours, a novel approach that identifies and aligns the most relevant visual and textual representations in a modular manner. Superior performance across various tasks demonstrates its capability to handle information misalignment and supports our identification theory.

2024

1. ICML

   Detecting and Identifying Selection Structure in Sequential Data

   Yujia Zheng, Zeyu Tang, Yiwen Qiu, Bernhard Schölkopf, and Kun Zhang

   In The Forty-First International Conference on Machine Learning, 2024.

   ABSTRACT arXiv PDF

   We argue that the selective inclusion of data points based on latent objectives is common in practical situations, such as music sequences. Since this selection process often distorts statistical analysis, previous work primarily views it as a bias to be corrected and proposes various methods to mitigate its effect. However, while controlling this bias is crucial, selection also offers an opportunity to provide a deeper insight into the hidden generation process, as it is a fundamental mechanism underlying what we observe. In particular, overlooking selection in sequential data can lead to an incomplete or overcomplicated inductive bias in modeling, such as assuming a universal autoregressive structure for all dependencies. Therefore, rather than merely viewing it as a bias, we explore the causal structure of selection in sequential data to delve deeper into the complete causal process. Specifically, we show that selection structure is identifiable without any parametric assumptions or interventional experiments. Moreover, even in cases where selection variables coexist with latent confounders, we still establish the nonparametric identifiability under appropriate structural conditions. Meanwhile, we also propose a provably correct algorithm to detect and identify selection structures as well as other types of dependencies. The framework has been validated empirically on both synthetic data and real-world music.

2. ICLRSpotlight

   Procedural Fairness Through Decoupling Objectionable Data Generating Components

   Zeyu Tang, Jialu Wang, Yang Liu, Peter Spirtes, and Kun Zhang

   In Twelfth International Conference on Learning Representations (preliminary version presented in NeurIPS 2023 AFT workshop), 2024.

   ABSTRACT arXiv PDF

   We reveal and address the frequently overlooked yet important issue of disguised procedural unfairness, namely, the potentially inadvertent alterations on the behavior of neutral (i.e., not problematic) aspects of data generating process, and/or the lack of procedural assurance of the greatest benefit of the least advantaged individuals. Inspired by John Rawls’s advocacy for pure procedural justice (Rawls, 1971; 2001), we view automated decision-making as a microcosm of social institutions, and consider how the data generating process itself can satisfy the requirements of procedural fairness. We propose a framework that decouples the objectionable data generating components from the neutral ones by utilizing reference points and the associated value instantiation rule. Our findings highlight the necessity of preventing disguised procedural unfairness, drawing attention not only to the objectionable data generating components that we aim to mitigate, but also more importantly, to the neutral components that we intend to keep unaffected.

3. R2FM @ ICLR

   Steering LLMs Towards Unbiased Responses: A Causality-Guided Debiasing Framework

   Jingling Li*, Zeyu Tang*, Xiaoyu Liu, Peter Spirtes, Kun Zhang, Liu Leqi, and Yang Liu

   ICLR 2024 Workshop on Reliable and Responsible Foundation Models (extended version titled "Prompting Fairness: Integrating Causality to Debias Large Language Models" is accepted to ICLR 2025), 2024.

   ABSTRACT arXiv PDF

   Large language models (LLMs) can easily generate biased and discriminative responses. As LLMs tap into consequential decision-making (e.g., hiring and healthcare), it is of crucial importance to develop strategies to mitigate these biases. This paper focuses on social bias, tackling the association between demographic information and LLM outputs. We propose a causality-guided debiasing framework that utilizes causal understandings of (1) the data-generating process of the training corpus fed to LLMs, and (2) the internal reasoning process of LLM inference, to guide the design of prompts for debiasing LLM outputs through selection mechanisms. Our framework unifies existing de-biasing prompting approaches such as inhibitive instructions and in-context contrastive examples, and sheds light on new ways of debiasing by encouraging bias-free reasoning. Our strong empirical performance on real-world datasets demonstrates that our framework provides principled guidelines on debiasing LLM outputs even with only the black-box access.

2023

1. ACM CSUR

   What-is and How-to for Fairness in Machine Learning: A Survey, Reflection, and Perspective

   Zeyu Tang, Jiji Zhang, and Kun Zhang

   ACM Computing Surveys, 2023.

   ABSTRACT arXiv PDF

   Algorithmic fairness has attracted increasing attention in the machine learning community. Various definitions are proposed in the literature, but the differences and connections among them are not clearly addressed. In this paper, we review and reflect on various fairness notions previously proposed in machine learning literature, and make an attempt to draw connections to arguments in moral and political philosophy, especially theories of justice. We also consider fairness inquiries from a dynamic perspective, and further consider the long-term impact that is induced by current prediction and decision. In light of the differences in the characterized fairness, we present a flowchart that encompasses implicit assumptions and expected outcomes of different types of fairness inquiries on the data generating process, on the predicted outcome, and on the induced impact, respectively. This paper demonstrates the importance of matching the mission (which kind of fairness one would like to enforce) and the means (which spectrum of fairness analysis is of interest, what is the appropriate analyzing scheme) to fulfill the intended purpose.

2. ICMLAward

   Model Transferability With Responsive Decision Subjects

   Yatong Chen, Zeyu Tang, Kun Zhang, and Yang Liu

   In The Fortieth International Conference on Machine Learning (preliminary version won the BEST PAPER AWARD in ICML 2022 AdvML workshop), 2023.

   ABSTRACT arXiv PDF

   This paper studies model transferability when human decision subjects respond to a deployed machine learning model. In our setting, an agent or a user corresponds to a sample (X,Y) drawn from a distribution D and will face a model h and its classification result h(X). Agents can modify X to adapt to h, which will incur a distribution shift on (X,Y). Therefore, when training h, the learner will need to consider the subsequently "induced" distribution when the output model is deployed. Our formulation is motivated by applications where the deployed machine learning models interact with human agents, and will ultimately face responsive and interactive data distributions. We formalize the discussions of the transferability of a model by studying how the model trained on the available source distribution (data) would translate to the performance on the induced domain. We provide both upper bounds for the performance gap due to the induced domain shift, as well as lower bound for the trade-offs that a classifier has to suffer on either the source training distribution or the induced target distribution. We provide further instantiated analysis for two popular domain adaptation settings with covariate shift and target shift.

3. ICLR

   Tier Balancing: Towards Dynamic Fairness over Underlying Causal Factors

   Zeyu Tang, Yatong Chen, Yang Liu, and Kun Zhang

   In The Eleventh International Conference on Learning Representations (preliminary version presented in NeurIPS 2022 AFCP workshop), 2023.

   ABSTRACT arXiv PDF

   The pursuit of long-term fairness involves the interplay between decision-making and the underlying data generating process. In this paper, through causal modeling with a directed acyclic graph (DAG) on the decision-distribution interplay, we investigate the possibility of achieving long-term fairness from a dynamic perspective. We propose Tier Balancing, a technically more challenging but more natural notion to achieve in the context of long-term, dynamic fairness analysis. Different from previous fairness notions that are defined purely on observed variables, our notion goes one step further, capturing behind-the-scenes situation changes on the unobserved latent causal factors that directly carry out the influence from the current decision to the future data distribution. Under the specified dynamics, we prove that in general one cannot achieve the long-term fairness goal only through one-step interventions. Furthermore, in the effort of approaching long-term fairness, we consider the mission of "getting closer to" the long-term fairness goal and present possibility and impossibility results accordingly.

2022

1. CLeaRSpotlight

   Attainability and Optimality: The Equalized Odds Fairness Revisited

   Zeyu Tang, and Kun Zhang

   In The First Conference on Causal Learning and Reasoning, 2022.

   ABSTRACT arXiv PDF

   Fairness of machine learning algorithms has been of increasing interest. In order to suppress or eliminate discrimination in prediction, various notions as well as approaches have been proposed to impose fairness. Given a notion of fairness, an essential problem is then whether or not it can always be attained, even if with an unlimited amount of data. This issue is, however, not well addressed yet. In this paper, focusing on the Equalized Odds notion of fairness, we consider the attainability of this criterion and, furthermore, if it is attainable, the optimality of the prediction performance under various settings. In particular, for prediction performed by a deterministic function of input features, we give conditions under which Equalized Odds can hold true; if the stochastic prediction is acceptable, we show that under mild assumptions, fair predictors can always be derived. For classification, we further prove that compared to enforcing fairness by post-processing, one can always benefit from exploiting all available features during training and get potentially better prediction performance while remaining fair. Moreover, while stochastic prediction can attain Equalized Odds with theoretical guarantees, we also discuss its limitation and potential negative social impacts.