Mohammad Mahdi Mehmanchi

Mohammad Mahdi Mehmanchi, Mahbod Nouri, Mohammad Sabokrou

This paper proposes a straightforward and cost-effective approach to assess whether a deep neural network (DNN) relies on the primary concepts of training samples or simply learns discriminative, yet simple and irrelevant features that can differentiate between classes. The paper highlights that DNNs, as discriminative classifiers, often find the simplest features to discriminate between classes, leading to a potential bias towards irrelevant features and sometimes missing generalization. While a generalization test is one way to evaluate a trained model's performance, it can be costly and may not cover all scenarios to ensure that the model has learned the primary concepts. Furthermore, even after conducting a generalization test, identifying bias in the model may not be possible. Here, the paper proposes a method that involves recovering samples from the parameters of the trained model and analyzing the reconstruction quality. We believe that if the model's weights are optimized to discriminate based on some features, these features will be reflected in the reconstructed samples. If the recovered samples contain the primary concepts of the training data, it can be concluded that the model has learned the essential and determining features. On the other hand, if the recovered samples contain irrelevant features, it can be concluded that the model is biased towards these features. The proposed method does not require any test or generalization samples, only the parameters of the trained model and the training data that lie on the margin. Our experiments demonstrate that the proposed method can determine whether the model has learned the desired features of the training data. The paper highlights that our understanding of how these models work is limited, and the proposed approach addresses this issue.

Seyedeh Fatemeh Razavi, Mohammad Mahdi Mehmanchi, Reshad Hosseini et al.

Using the intuition that out-of-distribution data have lower likelihoods, a common approach for out-of-distribution detection involves estimating the underlying data distribution. Normalizing flows are likelihood-based generative models providing a tractable density estimation via dimension-preserving invertible transformations. Conventional normalizing flows are prone to fail in out-of-distribution detection, because of the well-known curse of dimensionality problem of the likelihood-based models. To solve the problem of likelihood-based models, some works try to modify likelihood for example by incorporating a data complexity measure. We observed that these modifications are still insufficient. According to the manifold hypothesis, real-world data often lie on a low-dimensional manifold. Therefore, we proceed by estimating the density on a low-dimensional manifold and calculating a distance from the manifold as a measure for out-of-distribution detection. We propose a powerful criterion that combines this measure with the modified likelihood measure based on data complexity. Extensive experimental results show that incorporating manifold learning while accounting for the estimation of data complexity improves the out-of-distribution detection ability of normalizing flows. This improvement is achieved without modifying the model structure or using auxiliary out-of-distribution data during training.

Seyedeh Fatemeh Razavi, Mohammad Mahdi Mehmanchi, Reshad Hosseini et al.

Based on the manifold hypothesis, real-world data often lie on a low-dimensional manifold, while normalizing flows as a likelihood-based generative model are incapable of finding this manifold due to their structural constraints. So, one interesting question arises: $\textit{"Can we find sub-manifold(s) of data in normalizing flows and estimate the density of the data on the sub-manifold(s)?"}$. In this paper, we introduce two approaches, namely per-pixel penalized log-likelihood and hierarchical training, to answer the mentioned question. We propose a single-step method for joint manifold learning and density estimation by disentangling the transformed space obtained by normalizing flows to manifold and off-manifold parts. This is done by a per-pixel penalized likelihood function for learning a sub-manifold of the data. Normalizing flows assume the transformed data is Gaussianizationed, but this imposed assumption is not necessarily true, especially in high dimensions. To tackle this problem, a hierarchical training approach is employed to improve the density estimation on the sub-manifold. The results validate the superiority of the proposed methods in simultaneous manifold learning and density estimation using normalizing flows in terms of generated image quality and likelihood.