arXiv subjects

Weiwei Pan

Publications and source records attributed to Weiwei Pan.

At least 19 recordsLinked to original sources

arXiv2025

Rethinking LLM Bias Probing Using Lessons from the Social Sciences

The proliferation of LLM bias probes introduces three significant challenges: (1) we lack principled criteria for choosing appropriate probes, (2) we lack a system for reconciling conflicting results across probes, and (3) we lack formal frameworks for reasoning about when (and why) probe results will generalize to real user behavior. We address these challenges by systematizing LLM social bias probing using actionable insights from social sciences. We then introduce EcoLevels - a framework that helps (a) determine appropriate bias probes, (b) reconcile conflicting findings across probes, and (c) generate predictions about bias generalization. Overall, we ground our analysis in social science research because many LLM probes are direct applications of human probes, and these fields have faced similar challenges when studying social bias in humans. Based on our work, we suggest how the next generation of LLM bias probing can (and should) benefit from decades of social science research.

cs.CY↗

arXiv2024

Transparent Trade-offs between Properties of Explanations

When explaining black-box machine learning models, it's often important for explanations to have certain desirable properties. Most existing methods `encourage' desirable properties in their construction of explanations. In this work, we demonstrate that these forms of encouragement do not consistently create explanations with the properties that are supposedly being targeted. Moreover, they do not allow for any control over which properties are prioritized when different properties are at odds with each other. We propose to directly optimize explanations for desired properties. Our direct approach not only produces explanations with optimal properties more consistently but also empowers users to control trade-offs between different properties, allowing them to create explanations with exactly what is needed for a particular task.

cs.LG↗

arXiv2024

"ChatGPT, Don't Tell Me What to Do": Designing AI for Context Analysis in Humanitarian Frontline Negotiations

Frontline humanitarian negotiators are increasingly exploring ways to use AI tools in their workflows. However, current AI-tools in negotiation primarily focus on outcomes, neglecting crucial aspects of the negotiation process. Through iterative co-design with experienced frontline negotiators (n=32), we found that flexible tools that enable contextualizing cases and exploring options (with associated risks) are more effective than those providing direct recommendations of negotiation strategies. Surprisingly, negotiators demonstrated tolerance for occasional hallucinations and biases of AI. Our findings suggest that the design of AI-assisted negotiation tools should build on practitioners' existing practices, such as weighing different compromises and validating information with peers. This approach leverages negotiators' expertise while enhancing their decision-making capabilities. We call for technologists to learn from and collaborate closely with frontline negotiators, applying these insights to future AI designs and jointly developing professional guidelines for AI use in humanitarian negotiations.

cs.HC↗

arXiv2024

Is What You Ask For What You Get? Investigating Concept Associations in Text-to-Image Models

Text-to-image (T2I) models are increasingly used in impactful real-life applications. As such, there is a growing need to audit these models to ensure that they generate desirable, task-appropriate images. However, systematically inspecting the associations between prompts and generated content in a human-understandable way remains challenging. To address this, we propose Concept2Concept, a framework where we characterize conditional distributions of vision language models using interpretable concepts and metrics that can be defined in terms of these concepts. This characterization allows us to use our framework to audit models and prompt-datasets. To demonstrate, we investigate several case studies of conditional distributions of prompts, such as user-defined distributions or empirical, real-world distributions. Lastly, we implement Concept2Concept as an open-source interactive visualization tool to facilitate use by non-technical end-users. A demo is available at https://tinyurl.com/Concept2ConceptDemo.

cs.CV↗

arXiv2024

Inverse Reinforcement Learning with Multiple Planning Horizons

In this work, we study an inverse reinforcement learning (IRL) problem where the experts are planning under a shared reward function but with different, unknown planning horizons. Without the knowledge of discount factors, the reward function has a larger feasible solution set, which makes it harder for existing IRL approaches to identify a reward function. To overcome this challenge, we develop algorithms that can learn a global multi-agent reward function with agent-specific discount factors that reconstruct the expert policies. We characterize the feasible solution space of the reward function and discount factors for both algorithms and demonstrate the generalizability of the learned reward function across multiple domains.

cs.LG↗

arXiv2024

A Sim2Real Approach for Identifying Task-Relevant Properties in Interpretable Machine Learning

Explanations of an AI's function can assist human decision-makers, but the most useful explanation depends on the decision's context, referred to as the downstream task. User studies are necessary to determine the best explanations for each task. Unfortunately, testing every explanation and task combination is impractical, especially considering the many factors influencing human+AI collaboration beyond the explanation's content. This work leverages two insights to streamline finding the most effective explanation. First, explanations can be characterized by properties, such as faithfulness or complexity, which indicate if they contain the right information for the task. Second, we introduce XAIsim2real, a pipeline for running synthetic user studies. In our validation study, XAIsim2real accurately predicts user preferences across three tasks, making it a valuable tool for refining explanation choices before full studies. Additionally, it uncovers nuanced relationships, like how cognitive budget limits a user's engagement with complex explanations -- a trend confirmed with real users.

cs.HC↗

arXiv2024

Using Large Language Models for Humanitarian Frontline Negotiation: Opportunities and Considerations

Humanitarian negotiations in conflict zones, called \emph{frontline negotiation}, are often highly adversarial, complex, and high-risk. Several best-practices have emerged over the years that help negotiators extract insights from large datasets to navigate nuanced and rapidly evolving scenarios. Recent advances in large language models (LLMs) have sparked interest in the potential for AI to aid decision making in frontline negotiation. Through in-depth interviews with 13 experienced frontline negotiators, we identified their needs for AI-assisted case analysis and creativity support, as well as concerns surrounding confidentiality and model bias. We further explored the potential for AI augmentation of three standard tools used in frontline negotiation planning. We evaluated the quality and stability of our ChatGPT-based negotiation tools in the context of two real cases. Our findings highlight the potential for LLMs to enhance humanitarian negotiations and underscore the need for careful ethical and practical considerations.

cs.HC↗

arXiv2024

Towards Model-Agnostic Posterior Approximation for Fast and Accurate Variational Autoencoders

Inference for Variational Autoencoders (VAEs) consists of learning two models: (1) a generative model, which transforms a simple distribution over a latent space into the distribution over observed data, and (2) an inference model, which approximates the posterior of the latent codes given data. The two components are learned jointly via a lower bound to the generative model's log marginal likelihood. In early phases of joint training, the inference model poorly approximates the latent code posteriors. Recent work showed that this leads optimization to get stuck in local optima, negatively impacting the learned generative model. As such, recent work suggests ensuring a high-quality inference model via iterative training: maximizing the objective function relative to the inference model before every update to the generative model. Unfortunately, iterative training is inefficient, requiring heuristic criteria for reverting from iterative to joint training for speed. Here, we suggest an inference method that trains the generative and inference models independently. It approximates the posterior of the true model a priori; fixing this posterior approximation, we then maximize the lower bound relative to only the generative model. By conventional wisdom, this approach should rely on the true prior and likelihood of the true model to approximate its posterior (which are unknown). However, we show that we can compute a deterministic, model-agnostic posterior approximation (MAPA) of the true model's posterior. We then use MAPA to develop a proof-of-concept inference method. We present preliminary results on low-dimensional synthetic data that (1) MAPA captures the trend of the true posterior, and (2) our MAPA-based inference performs better density estimation with less computation than baselines. Lastly, we present a roadmap for scaling the MAPA-based inference method to high-dimensional data.

stat.ML↗

arXiv2024

Reinforcement Learning Interventions on Boundedly Rational Human Agents in Frictionful Tasks

Many important behavior changes are frictionful; they require individuals to expend effort over a long period with little immediate gratification. Here, an artificial intelligence (AI) agent can provide personalized interventions to help individuals stick to their goals. In these settings, the AI agent must personalize rapidly (before the individual disengages) and interpretably, to help us understand the behavioral interventions. In this paper, we introduce Behavior Model Reinforcement Learning (BMRL), a framework in which an AI agent intervenes on the parameters of a Markov Decision Process (MDP) belonging to a boundedly rational human agent. Our formulation of the human decision-maker as a planning agent allows us to attribute undesirable human policies (ones that do not lead to the goal) to their maladapted MDP parameters, such as an extremely low discount factor. Furthermore, we propose a class of tractable human models that captures fundamental behaviors in frictionful tasks. Introducing a notion of MDP equivalence specific to BMRL, we theoretically and empirically show that AI planning with our human models can lead to helpful policies on a wide range of more complex, ground-truth humans.

cs.AI↗

arXiv2023

Signature Activation: A Sparse Signal View for Holistic Saliency

The adoption of machine learning in healthcare calls for model transparency and explainability. In this work, we introduce Signature Activation, a saliency method that generates holistic and class-agnostic explanations for Convolutional Neural Network (CNN) outputs. Our method exploits the fact that certain kinds of medical images, such as angiograms, have clear foreground and background objects. We give theoretical explanation to justify our methods. We show the potential use of our method in clinical settings through evaluating its efficacy for aiding the detection of lesions in coronary angiograms.

cs.CV↗

arXiv2023

Why do universal adversarial attacks work on large language models?: Geometry might be the answer

Transformer based large language models with emergent capabilities are becoming increasingly ubiquitous in society. However, the task of understanding and interpreting their internal workings, in the context of adversarial attacks, remains largely unsolved. Gradient-based universal adversarial attacks have been shown to be highly effective on large language models and potentially dangerous due to their input-agnostic nature. This work presents a novel geometric perspective explaining universal adversarial attacks on large language models. By attacking the 117M parameter GPT-2 model, we find evidence indicating that universal adversarial triggers could be embedding vectors which merely approximate the semantic information in their adversarial training region. This hypothesis is supported by white-box model analysis comprising dimensionality reduction and similarity measurement of hidden representations. We believe this new geometric perspective on the underlying mechanism driving universal attacks could help us gain deeper insight into the internal workings and failure modes of LLMs, thus enabling their mitigation.

cs.LG↗

arXiv2023

SAP-sLDA: An Interpretable Interface for Exploring Unstructured Text

A common way to explore text corpora is through low-dimensional projections of the documents, where one hopes that thematically similar documents will be clustered together in the projected space. However, popular algorithms for dimensionality reduction of text corpora, like Latent Dirichlet Allocation (LDA), often produce projections that do not capture human notions of document similarity. We propose a semi-supervised human-in-the-loop LDA-based method for learning topics that preserve semantically meaningful relationships between documents in low-dimensional projections. On synthetic corpora, our method yields more interpretable projections than baseline methods with only a fraction of labels provided. On a real corpus, we obtain qualitatively similar results.

cs.CL↗

arXiv2023

The Unintended Consequences of Discount Regularization: Improving Regularization in Certainty Equivalence Reinforcement Learning

Discount regularization, using a shorter planning horizon when calculating the optimal policy, is a popular choice to restrict planning to a less complex set of policies when estimating an MDP from sparse or noisy data (Jiang et al., 2015). It is commonly understood that discount regularization functions by de-emphasizing or ignoring delayed effects. In this paper, we reveal an alternate view of discount regularization that exposes unintended consequences. We demonstrate that planning under a lower discount factor produces an identical optimal policy to planning using any prior on the transition matrix that has the same distribution for all states and actions. In fact, it functions like a prior with stronger regularization on state-action pairs with more transition data. This leads to poor performance when the transition matrix is estimated from data sets with uneven amounts of data across state-action pairs. Our equivalence theorem leads to an explicit formula to set regularization parameters locally for individual state-action pairs rather than globally. We demonstrate the failures of discount regularization and how we remedy them using our state-action-specific method across simple empirical examples as well as a medical cancer simulator.

cs.LG↗

arXiv2023

Multimodal N-of-1 trials: A Novel Personalized Healthcare Design

N-of-1 trials aim to estimate treatment effects on the individual level and can be applied to personalize a wide range of physical and digital interventions in mHealth. In this study, we propose and apply a framework for multimodal N-of-1 trials in order to allow the inclusion of health outcomes assessed through images, audio or videos. We illustrate the framework in a series of N-of-1 trials that investigate the effect of acne creams on acne severity assessed through pictures. For the analysis, we compare an expert-based manual labelling approach with different deep learning-based pipelines where in a first step, we train and fine-tune convolutional neural networks (CNN) on the images. Then, we use a linear mixed model on the scores obtained in the first step in order to test the effectiveness of the treatment. The results show that the CNN-based test on the images provides a similar conclusion as tests based on manual expert ratings of the images, and identifies a treatment effect in one individual. This illustrates that multimodal N-of-1 trials can provide a powerful way to identify individual treatment effects and can enable large-scale studies of a large variety of health outcomes that can be actively and passively assessed using technological advances in order to personalized health interventions.

stat.AP↗

arXiv2022

HubbardNet: Efficient Predictions of the Bose-Hubbard Model Spectrum with Deep Neural Networks

We present a deep neural network (DNN)-based model (HubbardNet) to variationally find the ground state and excited state wavefunctions of the one-dimensional and two-dimensional Bose-Hubbard model. Using this model for a square lattice with $M$ sites, we obtain the energy spectrum as an analytical function of the on-site Coulomb repulsion, $U$, and the total number of particles, $N$, from a single training. This approach bypasses the need to solve a new hamiltonian for each different set of values $(U,N)$. Using \texttt{HubbardNet}, we identify the two ground state phases of the Bose-Hubbard model (Mott insulator and superfluid). We show that the DNN-parametrized solutions are in excellent agreement with results from the exact diagonalization of the hamiltonian, and it outperforms exact diagonalization in terms of computational scaling. These advantages suggest that our model is promising for efficient and accurate computation of exact phase diagrams of many-body lattice hamiltonians.

cond-mat.str-el↗

arXiv2022

Modeling Mobile Health Users as Reinforcement Learning Agents

Mobile health (mHealth) technologies empower patients to adopt/maintain healthy behaviors in their daily lives, by providing interventions (e.g. push notifications) tailored to the user's needs. In these settings, without intervention, human decision making may be impaired (e.g. valuing near term pleasure over own long term goals). In this work, we formalize this relationship with a framework in which the user optimizes a (potentially impaired) Markov Decision Process (MDP) and the mHealth agent intervenes on the user's MDP parameters. We show that different types of impairments imply different types of optimal intervention. We also provide analytical and empirical explorations of these differences.

cs.LG↗

arXiv2022

An Empirical Analysis of the Advantages of Finite- v.s. Infinite-Width Bayesian Neural Networks

Comparing Bayesian neural networks (BNNs) with different widths is challenging because, as the width increases, multiple model properties change simultaneously, and, inference in the finite-width case is intractable. In this work, we empirically compare finite- and infinite-width BNNs, and provide quantitative and qualitative explanations for their performance difference. We find that when the model is mis-specified, increasing width can hurt BNN performance. In these cases, we provide evidence that finite-width BNNs generalize better partially due to the properties of their frequency spectrum that allows them to adapt under model mismatch.

stat.ML↗

arXiv2022

What Makes a Good Explanation?: A Harmonized View of Properties of Explanations

Interpretability provides a means for humans to verify aspects of machine learning (ML) models and empower human+ML teaming in situations where the task cannot be fully automated. Different contexts require explanations with different properties. For example, the kind of explanation required to determine if an early cardiac arrest warning system is ready to be integrated into a care setting is very different from the type of explanation required for a loan applicant to help determine the actions they might need to take to make their application successful. Unfortunately, there is a lack of standardization when it comes to properties of explanations: different papers may use the same term to mean different quantities, and different terms to mean the same quantity. This lack of a standardized terminology and categorization of the properties of ML explanations prevents us from both rigorously comparing interpretable machine learning methods and identifying what properties are needed in what contexts. In this work, we survey properties defined in interpretable machine learning papers, synthesize them based on what they actually measure, and describe the trade-offs between different formulations of these properties. In doing so, we enable more informed selection of task-appropriate formulations of explanation properties as well as standardization for future work in interpretable machine learning.

cs.LG↗