Neural networks are powerful function approximators, yet their ``black-box" nature often renders them opaque and difficult to interpret. While many post-hoc explanation methods exist, they typically fail to capture the underlying reasoning processes of the networks. A truly interpretable neural network would be trained similarly to conventional models using techniques such as backpropagation, but additionally provide insights into the learned input-output relationships. In this work, we introduce the concept of interpretability pipelineing, to incorporate multiple interpretability techniques to outperform each individual technique. To this end, we first evaluate several architectures that promise such interpretability, with a particular focus on two recent models selected for their potential to incorporate interpretability into standard neural network architectures while still leveraging backpropagation: the Growing Interpretable Neural Network (GINN) and Kolmogorov Arnold Networks (KAN). We analyze the limitations and strengths of each and introduce a novel interpretable neural network GINN-KAN that synthesizes the advantages of both models. When tested on the Feynman symbolic regression benchmark datasets, GINN-KAN outperforms both GINN and KAN. To highlight the capabilities and the generalizability of this approach, we position GINN-KAN as an alternative to conventional black-box networks in Physics-Informed Neural Networks (PINNs). We expect this to have far-reaching implications in the application of deep learning pipelines in the natural sciences. Our experiments with this interpretable PINN on 15 different partial differential equations demonstrate that GINN-KAN augmented PINNs outperform PINNs with black-box networks in solving differential equations and surpass the capabilities of both GINN and KAN.

Multivariate time-series forecasting (MTSF) stands as a compelling field within the machine learning community. Diverse neural network based methodologies deployed in MTSF applications have demonstrated commendable efficacy. Despite the advancements in model performance, comprehending the rationale behind the model's behavior remains an enigma. Our proposed model, the Neural ForeCasting Layer (NFCL), employs a straightforward amalgamation of neural networks. This uncomplicated integration ensures that each neural network contributes inputs and predictions independently, devoid of interference from other inputs. Consequently, our model facilitates a transparent explication of forecast results. This paper introduces NFCL along with its diverse extensions. Empirical findings underscore NFCL's superior performance compared to nine benchmark models across 15 available open datasets. Notably, NFCL not only surpasses competitors but also provides elucidation for its predictions. In addition, Rigorous experimentation involving diverse model structures bolsters the justification of NFCL's unique configuration.

Convolutional neural networks (CNNs) have shown exceptional performance for a range of medical imaging tasks. However, conventional CNNs are not able to explain their reasoning process, therefore limiting their adoption in clinical practice. In this work, we propose an inherently interpretable CNN for regression using similarity-based comparisons (INSightR-Net) and demonstrate our methods on the task of diabetic retinopathy grading. A prototype layer incorporated into the architecture enables visualization of the areas in the image that are most similar to learned prototypes. The final prediction is then intuitively modeled as a mean of prototype labels, weighted by the similarities. We achieved competitive prediction performance with our INSightR-Net compared to a ResNet baseline, showing that it is not necessary to compromise performance for interpretability. Furthermore, we quantified the quality of our explanations using sparsity and diversity, two concepts considered important for a good explanation, and demonstrated the effect of several parameters on the latent space embeddings.

In many areas, we have well-founded insights about causal structure that would be useful to bring into our trained models while still allowing them to learn in a data-driven fashion. To achieve this, we present the new method of interchange intervention training (IIT). In IIT, we (1) align variables in a causal model (e.g., a deterministic program or Bayesian network) with representations in a neural model and (2) train the neural model to match the counterfactual behavior of the causal model on a base input when aligned representations in both models are set to be the value they would be for a source input. IIT is fully differentiable, flexibly combines with other objectives, and guarantees that the target causal model is a causal abstraction of the neural model when its loss is zero. We evaluate IIT on a structural vision task (MNIST-PVR), a navigational language task (ReaSCAN), and a natural language inference task (MQNLI). We compare IIT against multi-task training objectives and data augmentation. In all our experiments, IIT achieves the best results and produces neural models that are more interpretable in the sense that they more successfully realize the target causal model.

This paper proposes two bottom-up interpretable neural network (NN) constructions for universal approximation, namely Triangularly-constructed NN (TNN) and Semi-Quantized Activation NN (SQANN). The notable properties are (1) resistance to catastrophic forgetting (2) existence of proof for arbitrarily high accuracies on training dataset (3) for an input \(x\), users can identify specific samples of training data whose activation ``fingerprints" are similar to that of \(x\)'s activations. Users can also identify samples that are out of distribution.

There are several approaches to rate machine learning models, two of them being accuracy and interpretability. A model with high accuracy is what we usually call a good model, it learned the relationship between the inputs X and outputs y well. If a model has high interpretability or explainability, we understand how the model makes a prediction and how we can influence this prediction by changing input features. While it is hard to say how the output of a deep neural network behaves when we increase or decrease a certain feature of the input, for a linear model it is extremely easy: if you increase the feature by one, the output increases by the coefficient of that feature. "There are interpretable models an there are well-performing models."

The increasing and ubiquitous use of machine learning (ML) algorithms in many technological [1], financial [2, 3] and medical applications [4] calls for an improved understanding of their inner working that is, calls for more interpretable algorithms. Indeed difficulties in understanding how neural networks operate constitute a major problem in sensitive applications such as self-driving vehicles or medical diagnosis, where errors from the machine could result in otherwise avoidable accidents and human losses. Actually, the impossibility to fully grasp the decision process undertaken by the network not only prevents humans from being able to supervise such decision and eventually correct it, but also hinders our ability to use these algorithms to better understand the problem under scrutiny, and to inspire new improved methods and approaches for solving it. Thus, the development and deployment of interpretable neural networks could represent an important step to improve the user trust and consequently to foster the adoption of Artificial Intelligence systems in common, everyday tasks [5, 6].