Recently, interpretable machine learning has re-explored concept bottleneck models (CBM), comprising step-by-step prediction of the high-level concepts from the raw features and the target variable from the predicted concepts. A compelling advantage of this model class is the user's ability to intervene on the predicted concept values, affecting the model's downstream output. In this work, we introduce a method to perform such concept-based interventions on already-trained neural networks, which are not interpretable by design, given an annotated validation set. Furthermore, we formalise the model's intervenability as a measure of the effectiveness of concept-based interventions and leverage this definition to fine-tune black-box models. Empirically, we explore the intervenability of black-box classifiers on synthetic tabular and natural image benchmarks. We demonstrate that fine-tuning improves intervention effectiveness and often yields better-calibrated predictions. To showcase the practical utility of the proposed techniques, we apply them to deep chest X-ray classifiers and show that fine-tuned black boxes can be as intervenable and more performant than CBMs.

Ricards Marcinkevics, Sonia Laguna, Moritz Vandenhirtz, Julia E. Vogt

Arxiv

24.01.2024

Link

We propose Tree Variational Autoencoder (TreeVAE), a new generative hierarchical clustering model that learns a flexible tree-based posterior distribution over latent variables. TreeVAE hierarchically divides samples according to their intrinsic characteristics, shedding light on hidden structures in the data. It adapts its architecture to discover the optimal tree for encoding dependencies between latent variables. The proposed tree-based generative architecture enables lightweight conditional inference and improves generative performance by utilizing specialized leaf decoders. We show that TreeVAE uncovers underlying clusters in the data and finds meaningful hierarchical relations between the different groups on a variety of datasets, including real-world imaging data. We present empirically that TreeVAE provides a more competitive log-likelihood lower bound than the sequential counterparts. Finally, due to its generative nature, TreeVAE is able to generate new samples from the discovered clusters via conditional sampling.

Laura Manduchi^*, Moritz Vandenhirtz^*, Alain Ryser, Julia E. Vogt
^* denotes shared first authorship

Spotlight at Neural Information Processing Systems, NeurIPS 2023

20.12.2023

LinkCode

Recently, interpretable machine learning has re-explored concept bottleneck models (CBM), comprising step-by-step prediction of the high-level concepts from the raw features and the target variable from the predicted concepts. A compelling advantage of this model class is the user's ability to intervene on the predicted concept values, consequently affecting the model's downstream output. In this work, we introduce a method to perform such concept-based interventions on already-trained neural networks, which are not interpretable by design. Furthermore, we formalise the model's intervenability as a measure of the effectiveness of concept-based interventions and leverage this definition to fine-tune black-box models. Empirically, we explore the intervenability of black-box classifiers on synthetic tabular and natural image benchmarks. We demonstrate that fine-tuning improves intervention effectiveness and often yields better-calibrated predictions. To showcase the practical utility of the proposed techniques, we apply them to chest X-ray classifiers and show that fine-tuned black boxes can be as intervenable and more performant than CBMs.

Ricards Marcinkevics, Sonia Laguna, Moritz Vandenhirtz, Julia E. Vogt

XAI in Action: Past, Present, and Future Applications, NeurIPS 2023

16.12.2023

Link

Prototype learning, a popular machine learning method designed for inherently interpretable decisions, leverages similarities to learned prototypes for classifying new data. While it is mainly applied in computer vision, in this work, we build upon prior research and further explore the extension of prototypical networks to natural language processing. We introduce a learned weighted similarity measure that enhances the similarity computation by focusing on informative dimensions of pre-trained sentence embeddings. Additionally, we propose a post-hoc explainability mechanism that extracts prediction-relevant words from both the prototype and input sentences. Finally, we empirically demonstrate that our proposed method not only improves predictive performance on the AG News and RT Polarity datasets over a previous prototype-based approach, but also improves the faithfulness of explanations compared to rationale-based recurrent convolutions.

Claudio Fanconi^*, Moritz Vandenhirtz^*, Severin Husmann, Julia E. Vogt
^* denotes shared first authorship

Conference on Empirical Methods in Natural Language Processing, EMNLP 2023

25.10.2023

LinkDOICode

We propose a new generative hierarchical clustering model that learns a flexible tree-based posterior distribution over latent variables. The proposed Tree Variational Autoencoder (TreeVAE) hierarchically divides samples according to their intrinsic characteristics, shedding light on hidden structures in the data. It adapts its architecture to discover the optimal tree for encoding dependencies between latent variables, improving generative performance. We show that TreeVAE uncovers underlying clusters in the data and finds meaningful hierarchical relations between the different groups on several datasets. Due to its generative nature, TreeVAE can generate new samples from the discovered clusters via conditional sampling.

Laura Manduchi^*, Moritz Vandenhirtz^*, Alain Ryser, Julia E. Vogt
^* denotes shared first authorship

ICML 2023 Workshop on Structured Probabilistic Inference & Generative Modeling

30.06.2023

LinkCode

We propose a new generative hierarchical clustering model that learns a flexible tree-based posterior distribution over latent variables. The proposed Tree Variational Autoencoder (TreeVAE) hierarchically divides samples according to their intrinsic characteristics, shedding light on hidden structures in the data. It adapts its architecture to discover the optimal tree for encoding dependencies between latent variables, improving generative performance. We show that TreeVAE uncovers underlying clusters in the data and finds meaningful hierarchical relations between the different groups on several datasets. Due to its generative nature, TreeVAE can generate new samples from the discovered clusters via conditional sampling.

Laura Manduchi^*, Moritz Vandenhirtz^*, Alain Ryser, Julia E. Vogt
^* denotes shared first authorship

ICML 2023 Workshop on Deployment Challenges for Generative AI

30.06.2023

LinkCode

Spurious correlations are everywhere. While humans often do not perceive them, neural networks are notorious for learning unwanted associations, also known as biases, instead of the underlying decision rule. As a result, practitioners are often unaware of the biased decision-making of their classifiers. Such a biased model based on spurious correlations might not generalize to unobserved data, leading to unintended, adverse consequences. We propose Signal is Harder (SiH), a variational-autoencoder-based method that simultaneously trains a biased and unbiased classifier using a novel, disentangling reweighting scheme inspired by the focal loss. Using the unbiased classifier, SiH matches or improves upon the performance of state-of-the-art debiasing methods. To improve the interpretability of our technique, we propose a perturbation scheme in the latent space for visualizing the bias that helps practitioners become aware of the sources of spurious correlations.

Moritz Vandenhirtz, Laura Manduchi, Ricards Marcinkevics, Julia E. Vogt

Domain Generalization Workshop, ICLR 2023

04.05.2023

LinkCode

Objective: To report the outcomes of active surveillance (AS) for low-risk prostate cancer (PCa) in a single-center cohort. Patients and Methods: This is a prospective, single-center, observational study. The cohort included all patients who underwent AS for PCa between December 1999 and December 2020 at our institution. Follow-up appointments (FU) ended in February 2021. Results: A total of 413 men were enrolled in the study, and 391 had at least one FU. Of those who followed up, 267 had PCa diagnosed by transrectal ultrasound (TRUS)-guided biopsy (T1c: 68.3%), while 124 were diagnosed after transurethral resection of the prostate (TURP) (T1a/b: 31.7%). Median FU was 46 months (IQR 25–90). Cancer specific survival was 99.7% and overall survival was 92.3%. Median reclassification time was 11.2 years. After 20 years, 25% of patients were reclassified within 4.58 years, 6.6% opted to switch to watchful waiting, 4.1% died, 17.4% were lost to FU, and 46.8% remained on AS. Those diagnosed by TRUS had a significantly higher reclassification rate than those diagnosed by TURP (p < 0.0001). Men diagnosed by targeted MRI/TRUS fusion biopsy tended to have a higher reclassification probability than those diagnosed by conventional template biopsies (p = 0.083). Conclusions: Our single-center cohort spanning over two decades revealed that AS remains a safe option for low-risk PCa even in the long term. Approximately half of AS enrollees will eventually require definitive treatment due to disease progression. Men with incidental prostate cancer were significantly less likely to have disease progression.

Sarah Hagmann, Venkat Ramakrishnan, Alexander Tamalunas, Marc Hofmann, Moritz Vandenhirtz, Silvan Vollmer, Jsmea Hug, Philipp Niggli, Antonio Nocito, Rahel A. Kubik-Huch, Kurt Lehmann, Lukas John Hefermehl

Cancers

12.01.2022

LinkDOI