Babak Nadjar Araabi

Sajad Movahedi, Melika Adabinejad, Ayyoob Imani et al.

Differentiable neural architecture search (DARTS) is a popular method for neural architecture search (NAS), which performs cell-search and utilizes continuous relaxation to improve the search efficiency via gradient-based optimization. The main shortcoming of DARTS is performance collapse, where the discovered architecture suffers from a pattern of declining quality during search. Performance collapse has become an important topic of research, with many methods trying to solve the issue through either regularization or fundamental changes to DARTS. However, the weight-sharing framework used for cell-search in DARTS and the convergence of architecture parameters has not been analyzed yet. In this paper, we provide a thorough and novel theoretical and empirical analysis on DARTS and its point of convergence. We show that DARTS suffers from a specific structural flaw due to its weight-sharing framework that limits the convergence of DARTS to saturation points of the softmax function. This point of convergence gives an unfair advantage to layers closer to the output in choosing the optimal architecture, causing performance collapse. We then propose two new regularization terms that aim to prevent performance collapse by harmonizing operation selection via aligning gradients of layers. Experimental results on six different search spaces and three different datasets show that our method ($Λ$-DARTS) does indeed prevent performance collapse, providing justification for our theoretical analysis and the proposed remedy.

Abdolmahdi Bagheri, Mahdi Dehshiri, Yamin Bagheri et al.

Neuroscientific studies aim to find an accurate and reliable brain Effective Connectome (EC). Although current EC discovery methods have contributed to our understanding of brain organization, their performances are severely constrained by the short sample size and poor temporal resolution of fMRI data, and high dimensionality of the brain connectome. By leveraging the DTI data as prior knowledge, we introduce two Bayesian causal discovery frameworks -- the Bayesian GOLEM (BGOLEM) and Bayesian FGES (BFGES) methods -- that offer significantly more accurate and reliable ECs and address the shortcomings of the existing causal discovery methods in discovering ECs based on only fMRI data. Through a series of simulation studies on synthetic and hybrid (DTI of the Human Connectome Project (HCP) subjects and synthetic fMRI) data, we demonstrate the effectiveness of the proposed methods in discovering EC. To numerically assess the improvement in the accuracy of ECs with our method on empirical data, we first introduce the Pseudo False Discovery Rate (PFDR) as a new computational accuracy metric for causal discovery in the brain. We show that our Bayesian methods achieve higher accuracy than traditional methods on HCP data. Additionally, we measure the reliability of discovered ECs using the Rogers-Tanimoto index for test-retest data and show that our Bayesian methods provide significantly more reproducible ECs than traditional methods. Overall, our study's numerical and graphical results highlight the potential for these frameworks to advance our understanding of brain function and organization significantly.

Mohammad Pasande, Reshad Hosseini, Babak Nadjar Araabi

Gaussian Mixture Models (GMMs) are one of the most potent parametric density models used extensively in many applications. Flexibly-tied factorization of the covariance matrices in GMMs is a powerful approach for coping with the challenges of common GMMs when faced with high-dimensional data and complex densities which often demand a large number of Gaussian components. However, the expectation-maximization algorithm for fitting flexibly-tied GMMs still encounters difficulties with streaming and very large dimensional data. To overcome these challenges, this paper suggests the use of first-order stochastic optimization algorithms. Specifically, we propose a new stochastic optimization algorithm on the manifold of orthogonal matrices. Through numerous empirical results on both synthetic and real datasets, we observe that stochastic optimization methods can outperform the expectation-maximization algorithm in terms of attaining better likelihood, needing fewer epochs for convergence, and consuming less time per each epoch.

Abdolmahdi Bagheri, Mohammad Pasande, Kevin Bello et al.

Understanding the complex mechanisms of the brain can be unraveled by extracting the Dynamic Effective Connectome (DEC). Recently, score-based Directed Acyclic Graph (DAG) discovery methods have shown significant improvements in extracting the causal structure and inferring effective connectivity. However, learning DEC through these methods still faces two main challenges: one with the fundamental impotence of high-dimensional dynamic DAG discovery methods and the other with the low quality of fMRI data. In this paper, we introduce Bayesian Dynamic DAG learning with M-matrices Acyclicity characterization (BDyMA) method to address the challenges in discovering DEC. The presented dynamic causal model enables us to discover direct feedback loop edges as well. Leveraging an unconstrained framework in the BDyMA method leads to more accurate results in detecting high-dimensional networks, achieving sparser outcomes, making it particularly suitable for extracting DEC. Additionally, the score function of the BDyMA method allows the incorporation of prior knowledge into the process of dynamic causal discovery which further enhances the accuracy of results. Comprehensive simulations on synthetic data and experiments on Human Connectome Project (HCP) data demonstrate that our method can handle both of the two main challenges, yielding more accurate and reliable DEC compared to state-of-the-art and traditional methods. Additionally, we investigate the trustworthiness of DTI data as prior knowledge for DEC discovery and show the improvements in DEC discovery when the DTI data is incorporated into the process.

Ali Javidani, Babak Nadjar Araabi, Mohammad Amin Sadeghi

This paper introduces a novel approach that integrates graph theory into self-supervised representation learning. Traditional methods focus on intra-instance variations generated by applying augmentations. However, they often overlook important inter-instance relationships. While our method retains the intra-instance property, it further captures inter-instance relationships by constructing k-nearest neighbor (KNN) graphs for both teacher and student streams during pretraining. In these graphs, nodes represent samples along with their latent representations. Edges encode the similarity between instances. Following pretraining, a representation refinement phase is performed. In this phase, Graph Neural Networks (GNNs) propagate messages not only among immediate neighbors but also across multiple hops, thereby enabling broader contextual integration. Experimental results on CIFAR-10, ImageNet-100, and ImageNet-1K demonstrate accuracy improvements of 7.3%, 3.2%, and 1.0%, respectively, over state-of-the-art methods. These results highlight the effectiveness of the proposed graph based mechanism. The code is publicly available at https://github.com/alijavidani/SSL-GraphNNCLR.

Ali Hedayatnia, Mostafa Tavassolipour, Babak Nadjar Araabi et al.

Randomized smoothing is a well-established method for achieving certified robustness against l2-adversarial perturbations. By incorporating a denoiser before the base classifier, pretrained classifiers can be seamlessly integrated into randomized smoothing without significant performance degradation. Among existing methods, Diffusion Denoised Smoothing - where a pretrained denoising diffusion model serves as the denoiser - has produced state-of-the-art results. However, we show that employing a denoising diffusion model introduces a covariate shift via misestimation of the added noise, ultimately degrading the smoothed classifier's performance. To address this issue, we propose a novel adversarial objective function focused on the added noise of the denoising diffusion model. This approach is inspired by our understanding of the origin of the covariate shift. Our goal is to train the base classifier to ensure it is robust against the covariate shift introduced by the denoiser. Our method significantly improves certified accuracy across three standard classification benchmarks - MNIST, CIFAR-10, and ImageNet - achieving new state-of-the-art performance in l2-adversarial perturbations. Our implementation is publicly available at https://github.com/ahedayat/Robustifying-DDS-Against-Covariate-Shift

Ali Javidani, Mohammad Amin Sadeghi, Babak Nadjar Araabi

Self-supervised visual representation learning traditionally focuses on image-level instance discrimination. Our study introduces an innovative, fine-grained dimension by integrating patch-level discrimination into these methodologies. This integration allows for the simultaneous analysis of local and global visual features, thereby enriching the quality of the learned representations. Initially, the original images undergo spatial augmentation. Subsequently, we employ a distinctive photometric patch-level augmentation, where each patch is individually augmented, independent from other patches within the same view. This approach generates a diverse training dataset with distinct color variations in each segment. The augmented images are then processed through a self-distillation learning framework, utilizing the Vision Transformer (ViT) as its backbone. The proposed method minimizes the representation distances across both image and patch levels to capture details from macro to micro perspectives. To this end, we present a simple yet effective patch-matching algorithm to find the corresponding patches across the augmented views. Thanks to the efficient structure of the patch-matching algorithm, our method reduces computational complexity compared to similar approaches. Consequently, we achieve an advanced understanding of the model without adding significant computational requirements. We have extensively pretrained our method on datasets of varied scales, such as Cifar10, ImageNet-100, and ImageNet-1K. It demonstrates superior performance over state-of-the-art self-supervised representation learning methods in image classification and downstream tasks, such as copy detection and image retrieval. The implementation of our method is accessible on GitHub.

Abdolmahdi Bagheri, Mahdi Dehshiri, Babak Nadjar Araabi et al.

In the investigation of any causal mechanisms, such as the brain's causal networks, the assumption of causal sufficiency plays a critical role. Notably, neglecting this assumption can result in significant errors, a fact that is often disregarded in the causal analysis of brain networks. In this study, we propose an algorithmic identification approach for determining essential exogenous nodes that satisfy the critical need for causal sufficiency to adhere to it in such inquiries. Our approach consists of three main steps: First, by capturing the essence of the Peter-Clark (PC) algorithm, we conduct independence tests for pairs of regions within a network, as well as for the same pairs conditioned on nodes from other networks. Next, we distinguish candidate confounders by analyzing the differences between the conditional and unconditional results, using the Kolmogorov-Smirnov test. Subsequently, we utilize Non-Factorized identifiable Variational Autoencoders (NF-iVAE) along with the Correlation Coefficient index (CCI) metric to identify the confounding variables within these candidate nodes. Applying our method to the Human Connectome Projects (HCP) movie-watching task data, we demonstrate that while interactions exist between dorsal and ventral regions, only dorsal regions serve as confounders for the visual networks, and vice versa. These findings align consistently with those resulting from the neuroscientific perspective. Finally, we show the reliability of our results by testing 30 independent runs for NF-iVAE initialization.

Farshad Sangari Abiz, Reshad Hosseini, Babak N. Araabi

Variational Autoencoders (VAEs) are powerful generative models for learning latent representations. Standard VAEs generate dispersed and unstructured latent spaces by utilizing all dimensions, which limits their interpretability, especially in high-dimensional spaces. To address this challenge, Variational Sparse Coding (VSC) introduces a spike-and-slab prior distribution, resulting in sparse latent representations for each input. These sparse representations, characterized by a limited number of active dimensions, are inherently more interpretable. Despite this advantage, VSC falls short in providing structured interpretations across samples within the same class. Intuitively, samples from the same class are expected to share similar attributes while allowing for variations in those attributes. This expectation should manifest as consistent patterns of active dimensions in their latent representations, but VSC does not enforce such consistency. In this paper, we propose a novel approach to enhance the latent space interpretability by ensuring that the active dimensions in the latent space are consistent across samples within the same class. To achieve this, we introduce a new loss function that encourages samples from the same class to share similar active dimensions. This alignment creates a more structured and interpretable latent space, where each shared dimension corresponds to a high-level concept, or "factor." Unlike existing disentanglement-based methods that primarily focus on global factors shared across all classes, our method captures both global and class-specific factors, thereby enhancing the utility and interpretability of latent representations.

Kiana Hooshanfar, Alireza Hosseini, Ahmad Kalhor et al.

Audio-visual saliency prediction aims to mimic human visual attention by identifying salient regions in videos through the integration of both visual and auditory information. Although visual-only approaches have significantly advanced, effectively incorporating auditory cues remains challenging due to complex spatio-temporal interactions and high computational demands. To address these challenges, we propose Dynamic Token Fusion Saliency (DFTSal), a novel audio-visual saliency prediction framework designed to balance accuracy with computational efficiency. Our approach features a multi-scale visual encoder equipped with two novel modules: the Learnable Token Enhancement Block (LTEB), which adaptively weights tokens to emphasize crucial saliency cues, and the Dynamic Learnable Token Fusion Block (DLTFB), which employs a shifting operation to reorganize and merge features, effectively capturing long-range dependencies and detailed spatial information. In parallel, an audio branch processes raw audio signals to extract meaningful auditory features. Both visual and audio features are integrated using our Adaptive Multimodal Fusion Block (AMFB), which employs local, global, and adaptive fusion streams for precise cross-modal fusion. The resulting fused features are processed by a hierarchical multi-decoder structure, producing accurate saliency maps. Extensive evaluations on six audio-visual benchmarks demonstrate that DFTSal achieves SOTA performance while maintaining computational efficiency.

Zahra Mousavi Kouzehkanan, Reshad Hosseini, Babak Nadjar Araabi

Many edge and contour detection algorithms give a soft-value as an output and the final binary map is commonly obtained by applying an optimal threshold. In this paper, we propose a novel method to detect image contours from the extracted edge segments of other algorithms. Our method is based on an undirected graphical model with the edge segments set as the vertices. The proposed energy functions are inspired by the surround modulation in the primary visual cortex that help suppressing texture noise. Our algorithm can improve extracting the binary map, because it considers other important factors such as connectivity, smoothness, and length of the contour beside the soft-values. Our quantitative and qualitative experimental results show the efficacy of the proposed method.

Saeed Masoudnia, Melika Kheirieh, Abdol-Hossein Vahabie et al.

Visual attention brings significant progress for Convolution Neural Networks (CNNs) in various applications. In this paper, object-based attention in human visual cortex inspires us to introduce a mechanism for modification of activations in feature maps of CNNs. In this mechanism, the activations of object locations are excited in feature maps. This mechanism is specifically inspired by attention-based gain modulation in object-based attention in brain. It facilitates figure-ground segregation in the visual cortex. Similar to brain, we use the idea to address two challenges in salient object detection: gathering object interior parts while segregation from background with concise boundaries. We implement the object-based attention in the U-net model using different architectures in the encoder parts, including AlexNet, VGG, and ResNet. The proposed method was examined on three benchmark datasets: HKU-IS, MSRB, and PASCAL-S. Experimental results showed that our inspired method could significantly improve the results in terms of mean absolute error and F-measure. The results also showed that our proposed method better captured not only the boundary but also the object interior. Thus, it can tackle the mentioned challenges.

Mohammad Mahdi Derakhshani, Saeed Masoudnia, Amir Hossein Shaker et al.

We present a simple and effective learning technique that significantly improves mAP of YOLO object detectors without compromising their speed. During network training, we carefully feed in localization information. We excite certain activations in order to help the network learn to better localize. In the later stages of training, we gradually reduce our assisted excitation to zero. We reached a new state-of-the-art in the speed-accuracy trade-off. Our technique improves the mAP of YOLOv2 by 3.8% and mAP of YOLOv3 by 2.2% on MSCOCO dataset.This technique is inspired from curriculum learning. It is simple and effective and it is applicable to most single-stage object detectors.

Ahmad Kalhor, Mohsen Saffar, Melika Kheirieh et al.

This paper introduces two straightforward, effective indices to evaluate the input data and the data flowing through layers of a feedforward deep neural network. For classification problems, the separation rate of target labels in the space of dataflow is explained as a key factor indicating the performance of designed layers in improving the generalization of the network. According to the explained concept, a shapeless distance-based evaluation index is proposed. Similarly, for regression problems, the smoothness rate of target outputs in the space of dataflow is explained as a key factor indicating the performance of designed layers in improving the generalization of the network. According to the explained smoothness concept, a shapeless distance-based smoothness index is proposed for regression problems. To consider more strictly concepts of separation and smoothness, their extended versions are introduced, and by interpreting a regression problem as a classification problem, it is shown that the separation and smoothness indices are related together. Through four case studies, the profits of using the introduced indices are shown. In the first case study, for classification and regression problems , the challenging of some known input datasets are compared respectively by the proposed separation and smoothness indices. In the second case study, the quality of dataflow is evaluated through layers of two pre-trained VGG 16 networks in classification of Cifar10 and Cifar100. In the third case study, it is shown that the correct classification rate and the separation index are almost equivalent through layers particularly while the serration index is increased. In the last case study, two multi-layer neural networks, which are designed for the prediction of Boston Housing price, are compared layer by layer by using the proposed smoothness index.

Saeed Masoudnia, Omid Mersa, Babak N. Araabi et al.

Offline Signature Verification (OSV) is a challenging pattern recognition task, especially in presence of skilled forgeries that are not available during training. This study aims to tackle its challenges and meet the substantial need for generalization for OSV by examining different loss functions for Convolutional Neural Network (CNN). We adopt our new approach to OSV by asking two questions: 1. which classification loss provides more generalization for feature learning in OSV? , and 2. How integration of different losses into a unified multi-loss function lead to an improved learning framework? These questions are studied based on analysis of three loss functions, including cross entropy, Cauchy-Schwarz divergence, and hinge loss. According to complementary features of these losses, we combine them into a dynamic multi-loss function and propose a novel ensemble framework for simultaneous use of them in CNN. Our proposed Multi-Loss Snapshot Ensemble (MLSE) consists of several sequential trials. In each trial, a dominant loss function is selected from the multi-loss set, and the remaining losses act as a regularizer. Different trials learn diverse representations for each input based on signature identification task. This multi-representation set is then employed for the verification task. An ensemble of SVMs is trained on these representations, and their decisions are finally combined according to the selection of most generalizable SVM for each user. We conducted two sets of experiments based on two different protocols of OSV, i.e., writer-dependent and writer-independent on three signature datasets: GPDS-Synthetic, MCYT, and UT-SIG. Based on the writer-dependent OSV protocol, we achieved substantial improvements over the best EERs in the literature. The results of the second set of experiments also confirmed the robustness to the arrival of new users enrolled in the OSV system.

Mina Alibeigi, Majid Nili Ahmadabadi, Babak Nadjar Araabi

Nowadays, robots become a companion in everyday life. To be well-accepted by humans, robots should efficiently understand meanings of their partners' motions and body language, and respond accordingly. Learning concepts by imitation brings them this ability in a user-friendly way. This paper presents a fast and robust model for Incremental Learning of Concepts by Imitation (ILoCI). In ILoCI, observed multimodal spatio-temporal demonstrations are incrementally abstracted and generalized based on both their perceptual and functional similarities during the imitation. In this method, perceptually similar demonstrations are abstracted by a dynamic model of mirror neuron system. An incremental method is proposed to learn their functional similarities through a limited number of interactions with the teacher. Learning all concepts together by the proposed memory rehearsal enables robot to utilize the common structural relations among concepts which not only expedites the learning process especially at the initial stages, but also improves the generalization ability and the robustness against discrepancies between observed demonstrations. Performance of ILoCI is assessed using standard LASA handwriting benchmark data set. The results show efficiency of ILoCI in concept acquisition, recognition and generation in addition to its robustness against variability in demonstrations.

Amir Soleimani, Kazim Fouladi, Babak N. Araabi

The pivotal role of datasets in signature verification systems motivates researchers to collect signature samples. Distinct characteristics of Persian signature demands for richer and culture-dependent offline signature datasets. This paper introduces a new and public Persian offline signature dataset, UTSig, that consists of 8280 images from 115 classes. Each class has 27 genuine signatures, 3 opposite-hand signatures, and 42 skilled forgeries made by 6 forgers. Compared with the other public datasets, UTSig has more samples, more classes, and more forgers. We considered various variables including signing period, writing instrument, signature box size, and number of observable samples for forgers in the data collection procedure. By careful examination of main characteristics of offline signature datasets, we observe that Persian signatures have fewer numbers of branch points and end points. We propose and evaluate four different training and test setups for UTSig. Results of our experiments show that training genuine samples along with opposite-hand samples and random forgeries can improve the performance in terms of equal error rate and minimum cost of log likelihood ratio.