Nicholas Kroeger

Nicholas Kroeger, Dan Ley, Satyapriya Krishna et al. · harvard

Recent advancements in Large Language Models (LLMs) have demonstrated exceptional capabilities in complex tasks like machine translation, commonsense reasoning, and language understanding. One of the primary reasons for the adaptability of LLMs in such diverse tasks is their in-context learning (ICL) capability, which allows them to perform well on new tasks by simply using a few task samples in the prompt. Despite their effectiveness in enhancing the performance of LLMs on diverse language and tabular tasks, these methods have not been thoroughly explored for their potential to generate post hoc explanations. In this work, we carry out one of the first explorations to analyze the effectiveness of LLMs in explaining other complex predictive models using ICL. To this end, we propose a novel framework, In-Context Explainers, comprising of three novel approaches that exploit the ICL capabilities of LLMs to explain the predictions made by other predictive models. We conduct extensive analysis with these approaches on real-world tabular and text datasets and demonstrate that LLMs are capable of explaining other predictive models similar to state-of-the-art post hoc explainers, opening up promising avenues for future research into LLM-based post hoc explanations of complex predictive models.

Nicholas M. Kroeger, Vincent Bindschaedler

While modern deep neural networks achieve impressive performance in vision tasks, they remain opaque in their decision processes, risking unwarranted trust, undetected biases and unexpected failures. We propose cluster paths, a post-hoc interpretability method that clusters activations at selected layers and represents each input as its sequence of cluster IDs. To assess these cluster paths, we introduce four metrics: path complexity (cognitive load), weighted-path purity (class alignment), decision-alignment faithfulness (predictive fidelity), and path agreement (stability under perturbations). In a spurious-cue CIFAR-10 experiment, cluster paths identify color-based shortcuts and collapse when the cue is removed. On a five-class CelebA hair-color task, they achieve 90% faithfulness and maintain 96% agreement under Gaussian noise without sacrificing accuracy. Scaling to a Vision Transformer pretrained on ImageNet, we extend cluster paths to concept paths derived from prompting a large language model on minimal path divergences. Finally, we show that cluster paths can serve as an effective out-of-distribution (OOD) detector, reliably flagging anomalous samples before the model generates over-confident predictions. Cluster paths uncover visual concepts, such as color palettes, textures, or object contexts, at multiple network depths, demonstrating that cluster paths scale to large vision models while generating concise and human-readable explanations.

Susan Meerdink, James Bocinsky, Alina Zare et al.

In remote sensing, it is often challenging to acquire or collect a large dataset that is accurately labeled. This difficulty is usually due to several issues, including but not limited to the study site's spatial area and accessibility, errors in the global positioning system (GPS), and mixed pixels caused by an image's spatial resolution. We propose an approach, with two variations, that estimates multiple target signatures from training samples with imprecise labels: Multi-Target Multiple Instance Adaptive Cosine Estimator (Multi-Target MI-ACE) and Multi-Target Multiple Instance Spectral Match Filter (Multi-Target MI-SMF). The proposed methods address the problems above by directly considering the multiple-instance, imprecisely labeled dataset. They learn a dictionary of target signatures that optimizes detection against a background using the Adaptive Cosine Estimator (ACE) and Spectral Match Filter (SMF). Experiments were conducted to test the proposed algorithms using a simulated hyperspectral dataset, the MUUFL Gulfport hyperspectral dataset collected over the University of Southern Mississippi-Gulfpark Campus, and the AVIRIS hyperspectral dataset collected over Santa Barbara County, California. Both simulated and real hyperspectral target detection experiments show the proposed algorithms are effective at learning target signatures and performing target detection.