Measuring the alignment between representations lets us understand similarities between the feature spaces of different models, such as Vision Transformers trained under diverse paradigms. However, traditional measures for representational alignment yield only scalar values that obscure how these spaces agree in terms of learned features. To address this, we combine alignment analysis with concept discovery, allowing a fine-grained breakdown of alignment into individual concepts. This approach reveals both universal concepts across models and each representation’s internal concept structure. We introduce a new definition of concepts as non-linear manifolds, hypothesizing they better capture the geometry of the feature space. A sanity check demonstrates the advantage of this manifold-based definition over linear baselines for concept-based alignment. Finally, our alignment analysis of four different ViTs shows that increased supervision tends to reduce semantic organization in learned representations.

The field of eXplainable Artificial Intelligence (XAI) has witnessed significant advancements in recent years. However, the majority of progress has been concentrated in the domains of computer vision and natural language processing. For time series data, where the input itself is often not interpretable, dedicated XAI research is scarce. In this work, we put forward a virtual inspection layer for transforming the time series to an interpretable representation and allows to propagate relevance attributions to this representation via local XAI methods. In this way, we extend the applicability of XAI methods to domains (e.g. speech) where the input is only interpretable after a transformation. In this work, we focus on the Fourier transformation which, is prominently applied in the preprocessing of time series, with Layer-wise Relevance Propagation (LRP) and refer to our method as DFT-LRP. We demonstrate the usefulness of DFT-LRP in various time series classification settings like audio and electronic health records. We showcase how DFT-LRP reveals differences in the classification strategies of models trained in different domains (e.g., time vs. frequency domain) or helps to discover how models act on spurious correlations in the data.

Deep neural networks are a promising tool for Audio Event Classification. In contrast to other data like natural images, there are many sensible and non-obvious representations for audio data, which could serve as input to these models. Due to their black-box nature, the effect of different input representations has so far mostly been investigated by measuring classification performance. In this work, we leverage eXplainable AI (XAI), to understand the underlying classification strategies of models trained on different input representations. Specifically, we compare two model architectures with regard to relevant input features used for Audio Event Detection: one directly processes the signal as the raw waveform, and the other takes its time-frequency spectrogram representation as input. We show how relevance heatmaps obtained via Layer-wise Relevance Propagation uncover representation-dependent decision strategies. With these insights, we can make a well-informed decision about the best model and input representation in terms of robustness and representativity. Further, we can test whether the model’s classification strategies align with human requirements.

PredDiff is a model-agnostic, local attribution method that is firmly rooted in probability theory. Its simple intuition is to measure prediction changes while marginalizing features. In this work, we clarify properties of PredDiff and its close connection to Shapley values. We stress important differences between classification and regression, which require a specific treatment within both formalisms. We extend PredDiff by introducing a new, well-founded measure for interaction effects between arbitrary feature subsets. The study of interaction effects represents an inevitable step towards a comprehensive understanding of black-box models and is particularly important for science applications. Equipped with our novel interaction measure, PredDiff is a promising model-agnostic approach for obtaining reliable, numerically inexpensive and theoretically sound attributions.

BACKGROUND: Immunotherapy is a promising route towards personalized cancer treatment. A key algorithmic challenge in this process is to decide if a given peptide (neoepitope) binds with the major histocompatibility complex (MHC). This is an active area of research and there are many MHC binding prediction algorithms that can predict the MHC binding affinity for a given peptide to a high degree of accuracy. However, most of the state-of-the-art approaches make use of complicated training and model selection procedures, are restricted to peptides of a certain length and/or rely on heuristics. RESULTS: We put forward USMPep, a simple recurrent neural network that reaches state-of-the-art approaches on MHC class I binding prediction with a single, generic architecture and even a single set of hyperparameters both on IEDB benchmark datasets and on the very recent HPV dataset. Moreover, the algorithm is competitive for a single model trained from scratch, while ensembling multiple regressors and language model pretraining can still slightly improve the performance. The direct application of the approach to MHC class II binding prediction shows a solid performance despite of limited training data. CONCLUSIONS: We demonstrate that competitive performance in MHC binding affinity prediction can be reached with a standard architecture and training procedure without relying on any heuristics.