Edward Chen

Edward Chen, Sanmi Koyejo, Carlos Guestrin

Large language model (LLM) alignment relies on complex reward signals that often obscure the specific behaviors being incentivized, creating critical risks of misalignment and reward hacking. Existing interpretation methods typically rely on pre-defined rubrics, risking the omission of "unknown unknowns", or fail to identify objectives that comprehensively cover and are causal to the model behavior. To address these limitations, we introduce Obj-Disco, a framework that automatically decomposes an alignment reward signal into a sparse, weighted combination of human-interpretable natural language objectives. Our approach utilizes an iterative greedy algorithm to analyze behavioral changes across training checkpoints, identifying and validating candidate objectives that best explain the residual reward signal. Extensive evaluations across diverse tasks, model sizes, and alignment algorithms demonstrate the framework's robustness. Experiments with popular open-source reward models show that the framework consistently captures > 90% of reward behavior, a finding further corroborated by human evaluation. Additionally, a case study on alignment with an open-source reward model reveals that Obj-Disco can successfully identify latent misaligned incentives that emerge alongside intended behaviors. Our work provides a crucial tool for uncovering the implicit objectives in LLM alignment, paving the way for more transparent and safer AI development.

Edward Chen, Linyu Lin, Nam T. Dinh

The use of machine learning (ML) model as digital-twins for reduced-order-modeling (ROM) in lieu of system codes has grown traction over the past few years. However, due to the complex and non-linear nature of nuclear reactor transients as well as the large range of tasks required, it is infeasible for a single ML model to generalize across all tasks. In this paper, we incorporate issue specific digital-twin ML models with ensembles to enhance the prediction outcome. The ensemble also utilizes an indirect probabilistic tracking method of surrogate state variables to produce accurate predictions of unobservable safety goals. The unique method named Ensemble Diagnostic Digital-twin Modeling (EDDM) can select not only the most appropriate predictions from the incorporated diagnostic digital-twin models but can also reduce generalization error associated with training as opposed to single models.

Chenxiao Gao, Edward Chen, Tianyi Chen et al.

Thanks to their remarkable flexibility, diffusion models and flow models have emerged as promising candidates for policy representation. However, efficient reinforcement learning (RL) upon these policies remains a challenge due to the lack of explicit log-probabilities for vanilla policy gradient estimators. While numerous attempts have been proposed to address this, the field lacks a unified perspective to reconcile these seemingly disparate methods, thus hampering ongoing development. In this paper, we bridge this gap by introducing a comprehensive taxonomy for RL algorithms with diffusion/flow policies. To support reproducibility and agile prototyping, we introduce a modular, JAX-based open-source codebase that leverages JIT-compilation for high-throughput training. Finally, we provide systematic and standardized benchmarks across Gym-Locomotion, DeepMind Control Suite, and IsaacLab, offering a rigorous side-by-side comparison of diffusion-based methods and guidance for practitioners to choose proper algorithms based on the application. Our work establishes a clear foundation for understanding and algorithm design, a high-efficiency toolkit for future research in the field, and an algorithmic guideline for practitioners in generative models and robotics. Our code is available at https://github.com/typoverflow/flow-rl.

Edward Chen, Han Bao, Nam Dinh

In recent years, the field of data-driven neural network-based machine learning (ML) algorithms has grown significantly and spurred research in its applicability to instrumentation and control systems. While they are promising in operational contexts, the trustworthiness of such algorithms is not adequately assessed. Failures of ML-integrated systems are poorly understood; the lack of comprehensive risk modeling can degrade the trustworthiness of these systems. In recent reports by the National Institute for Standards and Technology, trustworthiness in ML is a critical barrier to adoption and will play a vital role in intelligent systems' safe and accountable operation. Thus, in this work, we demonstrate a real-time model-agnostic method to evaluate the relative reliability of ML predictions by incorporating out-of-distribution detection on the training dataset. It is well documented that ML algorithms excel at interpolation (or near-interpolation) tasks but significantly degrade at extrapolation. This occurs when new samples are "far" from training samples. The method, referred to as the Laplacian distributed decay for reliability (LADDR), determines the difference between the operational and training datasets, which is used to calculate a prediction's relative reliability. LADDR is demonstrated on a feedforward neural network-based model used to predict safety significant factors during different loss-of-flow transients. LADDR is intended as a "data supervisor" and determines the appropriateness of well-trained ML models in the context of operational conditions. Ultimately, LADDR illustrates how training data can be used as evidence to support the trustworthiness of ML predictions when utilized for conventional interpolation tasks.

Edward Chen, Natalie Dullerud, Thomas Niedermayr et al.

Countless science and engineering applications in multi-objective optimization (MOO) necessitate that decision-makers (DMs) select a Pareto-optimal solution which aligns with their preferences. Evaluating individual solutions is often expensive, necessitating cost-sensitive optimization techniques. Due to competing objectives, the space of trade-offs is also expansive -- thus, examining the full Pareto frontier may prove overwhelming to a DM. Such real-world settings generally have loosely-defined and context-specific desirable regions for each objective function that can aid in constraining the search over the Pareto frontier. We introduce a novel conceptual framework that operationalizes these priors using soft-hard functions, SHFs, which allow for the DM to intuitively impose soft and hard bounds on each objective -- which has been lacking in previous MOO frameworks. Leveraging a novel minimax formulation for Pareto frontier sampling, we propose a two-step process for obtaining a compact set of Pareto-optimal points which respect the user-defined soft and hard bounds: (1) densely sample the Pareto frontier using Bayesian optimization, and (2) sparsify the selected set to surface to the user, using robust submodular function optimization. We prove that (2) obtains the optimal compact Pareto-optimal set of points from (1). We further show that many practical problems fit within the SHF framework and provide extensive empirical validation on diverse domains, including brachytherapy, engineering design, and large language model personalization. Specifically, for brachytherapy, our approach returns a compact set of points with over 3% greater SHF-defined utility than the next best approach. Among the other diverse experiments, our approach consistently leads in utility, allowing the DM to reach >99% of their maximum possible desired utility within validation of 5 points.

Edward Chen, Sang T. Truong, Natalie Dullerud et al.

High-stakes decision-making involves navigating multiple competing objectives with expensive evaluations. For instance, in brachytherapy, clinicians must balance maximizing tumor coverage (e.g., an aspirational target or soft bound of >95% coverage) against strict organ dose limits (e.g., a non-negotiable hard bound of <601 cGy to the bladder), with each plan evaluation being resource-intensive. Selecting Pareto-optimal solutions that match implicit preferences is challenging, as exhaustive Pareto frontier exploration is computationally and cognitively prohibitive, necessitating interactive frameworks to guide users. While decision-makers (DMs) often possess domain knowledge to narrow the search via such soft-hard bounds, current methods often lack systematic approaches to iteratively refine these multi-faceted preference structures. Critically, DMs must trust their final decision, confident they haven't missed superior alternatives; this trust is paramount in high-consequence scenarios. We present Active-MoSH, an interactive local-global framework designed for this process. Its local component integrates soft-hard bounds with probabilistic preference learning, maintaining distributions over DM preferences and bounds for adaptive Pareto subset refinement. This is guided by an active sampling strategy optimizing exploration-exploitation while minimizing cognitive burden. To build DM trust, Active-MoSH's global component, T-MoSH, leverages multi-objective sensitivity analysis to identify potentially overlooked, high-value points beyond immediate feedback. We demonstrate Active-MoSH's performance benefits through diverse synthetic and real-world applications. A user study on AI-generated image selection further validates our hypotheses regarding the framework's ability to improve convergence, enhance DM trust, and provide expressive preference articulation, enabling more effective DMs.

Gautam Rajendrakumar Gare, Wanwen Chen, Alex Ling Yu Hung et al.

In this paper, we study the significance of the pleura and adipose tissue in lung ultrasound AI analysis. We highlight their more prominent appearance when using high-frequency linear (HFL) instead of curvilinear ultrasound probes, showing HFL reveals better pleura detail. We compare the diagnostic utility of the pleura and adipose tissue using an HFL ultrasound probe. Masking the adipose tissue during training and inference (while retaining the pleural line and Merlin's space artifacts such as A-lines and B-lines) improved the AI model's diagnostic accuracy.

Alex Ling Yu Hung, Edward Chen, John Galeotti

Ultrasound image quality has continually been improving. However, when needles or other metallic objects are operating inside the tissue, the resulting reverberation artifacts can severely corrupt the surrounding image quality. Such effects are challenging for existing computer vision algorithms for medical image analysis. Needle reverberation artifacts can be hard to identify at times and affect various pixel values to different degrees. The boundaries of such artifacts are ambiguous, leading to disagreement among human experts labeling the artifacts. We propose a weakly- and semi-supervised, probabilistic needle-and-reverberation-artifact segmentation algorithm to separate the desired tissue-based pixel values from the superimposed artifacts. Our method models the intensity decay of artifact intensities and is designed to minimize the human labeling error. We demonstrate the applicability of the approach and compare it against other segmentation algorithms. Our method is capable of differentiating between the reverberations from artifact-free patches as well as of modeling the intensity fall-off in the artifacts. Our method matches state-of-the-art artifact segmentation performance and sets a new standard in estimating the per-pixel contributions of artifact vs underlying anatomy, especially in the immediately adjacent regions between reverberation lines. Our algorithm is also able to improve the performance downstream image analysis algorithms.

Edward Chen, Tejas Sudharshan Mathai, Vinit Sarode et al.

Identifying landmarks in the femoral area is crucial for ultrasound (US) -based robot-guided catheter insertion, and their presentation varies when imaged with different scanners. As such, the performance of past deep learning-based approaches is also narrowly limited to the training data distribution; this can be circumvented by fine-tuning all or part of the model, yet the effects of fine-tuning are seldom discussed. In this work, we study the US-based segmentation of multiple classes through transfer learning by fine-tuning different contiguous blocks within the model, and evaluating on a gamut of US data from different scanners and settings. We propose a simple method for predicting generalization on unseen datasets and observe statistically significant differences between the fine-tuning methods while working towards domain generalization.

Edward Chen

This paper introduces the YouTube-8M Video Understanding Challenge hosted as a Kaggle competition and also describes my approach to experimenting with various models. For each of my experiments, I provide the score result as well as possible improvements to be made. Towards the end of the paper, I discuss the various ensemble learning techniques that I applied on the dataset which significantly boosted my overall competition score. At last, I discuss the exciting future of video understanding research and also the many applications that such research could significantly improve.