Mohsen Fayyaz

Amirhossein Kazerouni, Ehsan Khodapanah Aghdam, Moein Heidari et al.

Denoising diffusion models, a class of generative models, have garnered immense interest lately in various deep-learning problems. A diffusion probabilistic model defines a forward diffusion stage where the input data is gradually perturbed over several steps by adding Gaussian noise and then learns to reverse the diffusion process to retrieve the desired noise-free data from noisy data samples. Diffusion models are widely appreciated for their strong mode coverage and quality of the generated samples despite their known computational burdens. Capitalizing on the advances in computer vision, the field of medical imaging has also observed a growing interest in diffusion models. To help the researcher navigate this profusion, this survey intends to provide a comprehensive overview of diffusion models in the discipline of medical image analysis. Specifically, we introduce the solid theoretical foundation and fundamental concepts behind diffusion models and the three generic diffusion modelling frameworks: diffusion probabilistic models, noise-conditioned score networks, and stochastic differential equations. Then, we provide a systematic taxonomy of diffusion models in the medical domain and propose a multi-perspective categorization based on their application, imaging modality, organ of interest, and algorithms. To this end, we cover extensive applications of diffusion models in the medical domain. Furthermore, we emphasize the practical use case of some selected approaches, and then we discuss the limitations of the diffusion models in the medical domain and propose several directions to fulfill the demands of this field. Finally, we gather the overviewed studies with their available open-source implementations at https://github.com/amirhossein-kz/Awesome-Diffusion-Models-in-Medical-Imaging.

Ali Modarressi, Mohsen Fayyaz, Yadollah Yaghoobzadeh et al.

There has been a growing interest in interpreting the underlying dynamics of Transformers. While self-attention patterns were initially deemed as the primary option, recent studies have shown that integrating other components can yield more accurate explanations. This paper introduces a novel token attribution analysis method that incorporates all the components in the encoder block and aggregates this throughout layers. Through extensive quantitative and qualitative experiments, we demonstrate that our method can produce faithful and meaningful global token attributions. Our experiments reveal that incorporating almost every encoder component results in increasingly more accurate analysis in both local (single layer) and global (the whole model) settings. Our global attribution analysis significantly outperforms previous methods on various tasks regarding correlation with gradient-based saliency scores. Our code is freely available at https://github.com/mohsenfayyaz/GlobEnc.

Ali Modarressi, Mohsen Fayyaz, Ehsan Aghazadeh et al.

An emerging solution for explaining Transformer-based models is to use vector-based analysis on how the representations are formed. However, providing a faithful vector-based explanation for a multi-layer model could be challenging in three aspects: (1) Incorporating all components into the analysis, (2) Aggregating the layer dynamics to determine the information flow and mixture throughout the entire model, and (3) Identifying the connection between the vector-based analysis and the model's predictions. In this paper, we present DecompX to tackle these challenges. DecompX is based on the construction of decomposed token representations and their successive propagation throughout the model without mixing them in between layers. Additionally, our proposal provides multiple advantages over existing solutions for its inclusion of all encoder components (especially nonlinear feed-forward networks) and the classification head. The former allows acquiring precise vectors while the latter transforms the decomposition into meaningful prediction-based values, eliminating the need for norm- or summation-based vector aggregation. According to the standard faithfulness evaluations, DecompX consistently outperforms existing gradient-based and vector-based approaches on various datasets. Our code is available at https://github.com/mohsenfayyaz/DecompX.

Tanveer Hannan, Dimitrios Mallios, Parth Pathak et al.

Large Vision-Language Models (LVLMs) have demonstrated strong multimodal reasoning capabilities on long and complex documents. However, their high memory footprint makes them impractical for deployment on resource-constrained edge devices. We present DocSLM, an efficient Small Vision-Language Model designed for long-document understanding under constrained memory resources. DocSLM incorporates a Hierarchical Multimodal Compressor that jointly encodes visual, textual, and layout information from each page into a fixed-length sequence, greatly reducing memory consumption while preserving both local and global semantics. To enable scalable processing over arbitrarily long inputs, we introduce a Streaming Abstention mechanism that operates on document segments sequentially and filters low-confidence responses using an entropy-based uncertainty calibrator. Across multiple long multimodal document benchmarks, DocSLM matches or surpasses state-of-the-art methods while using 82\% fewer visual tokens, 75\% fewer parameters, and 71\% lower latency, delivering reliable multimodal document understanding on lightweight edge devices. Code is available in the supplementary material.

Mohsen Fayyaz, Ehsan Aghazadeh, Ali Modarressi et al.

Current pre-trained language models rely on large datasets for achieving state-of-the-art performance. However, past research has shown that not all examples in a dataset are equally important during training. In fact, it is sometimes possible to prune a considerable fraction of the training set while maintaining the test performance. Established on standard vision benchmarks, two gradient-based scoring metrics for finding important examples are GraNd and its estimated version, EL2N. In this work, we employ these two metrics for the first time in NLP. We demonstrate that these metrics need to be computed after at least one epoch of fine-tuning and they are not reliable in early steps. Furthermore, we show that by pruning a small portion of the examples with the highest GraNd/EL2N scores, we can not only preserve the test accuracy, but also surpass it. This paper details adjustments and implementation choices which enable GraNd and EL2N to be applied to NLP.

Ehsan Aghazadeh, Mohsen Fayyaz, Yadollah Yaghoobzadeh

Human languages are full of metaphorical expressions. Metaphors help people understand the world by connecting new concepts and domains to more familiar ones. Large pre-trained language models (PLMs) are therefore assumed to encode metaphorical knowledge useful for NLP systems. In this paper, we investigate this hypothesis for PLMs, by probing metaphoricity information in their encodings, and by measuring the cross-lingual and cross-dataset generalization of this information. We present studies in multiple metaphor detection datasets and in four languages (i.e., English, Spanish, Russian, and Farsi). Our extensive experiments suggest that contextual representations in PLMs do encode metaphorical knowledge, and mostly in their middle layers. The knowledge is transferable between languages and datasets, especially when the annotation is consistent across training and testing sets. Our findings give helpful insights for both cognitive and NLP scientists.

Ali Modarressi, Abdullatif Köksal, Ayyoob Imani et al.

While current large language models (LLMs) perform well on many knowledge-related tasks, they are limited by relying on their parameters as an implicit storage mechanism. As a result, they struggle with memorizing rare events and with updating their memory as facts change over time. In addition, the uninterpretable nature of parametric memory makes it challenging to prevent hallucination. Model editing and augmenting LLMs with parameters specialized for memory are only partial solutions. In this paper, we introduce MemLLM, a novel method of enhancing LLMs by integrating a structured and explicit read-and-write memory module. MemLLM tackles the aforementioned challenges by enabling dynamic interaction with the memory and improving the LLM's capabilities in using stored knowledge. Our experiments indicate that MemLLM enhances the LLM's performance and interpretability, in language modeling in general and knowledge-intensive tasks in particular. We see MemLLM as an important step towards making LLMs more grounded and factual through memory augmentation. The project repository is publicly available at https://github.com/amodaresi/MemLLM

Emad Bahrami, Olga Zatsarynna, Parth Pathak et al.

We introduce STRIVE (SpatioTemporal Reinforcement with Importance-aware Variant Exploration), a structured reinforcement learning framework for video question answering. While group-based policy optimization methods have shown promise in large multimodal models, they often suffer from low reward variance when responses exhibit similar correctness, leading to weak or unstable advantage estimates. STRIVE addresses this limitation by constructing multiple spatiotemporal variants of each input video and performing joint normalization across both textual generations and visual variants. By expanding group comparisons beyond linguistic diversity to structured visual perturbations, STRIVE enriches reward signals and promotes more stable and informative policy updates. To ensure exploration remains semantically grounded, we introduce an importance-aware sampling mechanism that prioritizes frames most relevant to the input question while preserving temporal coverage. This design encourages robust reasoning across complementary visual perspectives rather than overfitting to a single spatiotemporal configuration. Experiments on six challenging video reasoning benchmarks including VideoMME, TempCompass, VideoMMMU, MMVU, VSI-Bench, and PerceptionTest demonstrate consistent improvements over strong reinforcement learning baselines across multiple large multimodal models. Our results highlight the role of structured spatiotemporal exploration as a principled mechanism for stabilizing multimodal reinforcement learning and improving video reasoning performance.

Mohsen Fayyaz, Ali Modarressi, Hinrich Schuetze et al.

Dense retrieval models are commonly used in Information Retrieval (IR) applications, such as Retrieval-Augmented Generation (RAG). Since they often serve as the first step in these systems, their robustness is critical to avoid downstream failures. In this work, we repurpose a relation extraction dataset (e.g., Re-DocRED) to design controlled experiments that quantify the impact of heuristic biases, such as a preference for shorter documents, on retrievers like Dragon+ and Contriever. We uncover major vulnerabilities, showing retrievers favor shorter documents, early positions, repeated entities, and literal matches, all while ignoring the answer's presence! Notably, when multiple biases combine, models exhibit catastrophic performance degradation, selecting the answer-containing document in less than 10% of cases over a synthetic biased document without the answer. Furthermore, we show that these biases have direct consequences for downstream applications like RAG, where retrieval-preferred documents can mislead LLMs, resulting in a 34% performance drop than providing no documents at all. https://huggingface.co/datasets/mohsenfayyaz/ColDeR

Saeid Ghafouri, Mohsen Fayyaz, Xiangchen Li et al.

Real-time multi-label video classification on embedded devices is constrained by limited compute and energy budgets. Yet, video streams exhibit structural properties such as label sparsity, temporal continuity, and label co-occurrence that can be leveraged for more efficient inference. We introduce Polymorph, a context-aware framework that activates a minimal set of lightweight Low Rank Adapters (LoRA) per frame. Each adapter specializes in a subset of classes derived from co-occurrence patterns and is implemented as a LoRA weight over a shared backbone. At runtime, Polymorph dynamically selects and composes only the adapters needed to cover the active labels, avoiding full-model switching and weight merging. This modular strategy improves scalability while reducing latency and energy overhead. Polymorph achieves 40% lower energy consumption and improves mAP by 9 points over strong baselines on the TAO dataset. Polymorph is open source at https://github.com/inference-serving/polymorph/.

Seyyed Hossein Hasanpour, Mohammad Rouhani, Mohsen Fayyaz et al.

Major winning Convolutional Neural Networks (CNNs), such as VGGNet, ResNet, DenseNet, \etc, include tens to hundreds of millions of parameters, which impose considerable computation and memory overheads. This limits their practical usage in training and optimizing for real-world applications. On the contrary, light-weight architectures, such as SqueezeNet, are being proposed to address this issue. However, they mainly suffer from low accuracy, as they have compromised between the processing power and efficiency. These inefficiencies mostly stem from following an ad-hoc designing procedure. In this work, we discuss and propose several crucial design principles for an efficient architecture design and elaborate intuitions concerning different aspects of the design procedure. Furthermore, we introduce a new layer called {\it SAF-pooling} to improve the generalization power of the network while keeping it simple by choosing best features. Based on such principles, we propose a simple architecture called {\it SimpNet}. We empirically show that SimpNet provides a good trade-off between the computation/memory efficiency and the accuracy solely based on these primitive but crucial principles. SimpNet outperforms the deeper and more complex architectures such as VGGNet, ResNet, WideResidualNet \etc, on several well-known benchmarks, while having 2 to 25 times fewer number of parameters and operations. We obtain state-of-the-art results (in terms of a balance between the accuracy and the number of involved parameters) on standard datasets, such as CIFAR10, CIFAR100, MNIST and SVHN. The implementations are available at \href{url}{https://github.com/Coderx7/SimpNet}.

Seyyed Hossein Hasanpour, Mohammad Rouhani, Mohsen Fayyaz et al.

Major winning Convolutional Neural Networks (CNNs), such as AlexNet, VGGNet, ResNet, GoogleNet, include tens to hundreds of millions of parameters, which impose considerable computation and memory overhead. This limits their practical use for training, optimization and memory efficiency. On the contrary, light-weight architectures, being proposed to address this issue, mainly suffer from low accuracy. These inefficiencies mostly stem from following an ad hoc procedure. We propose a simple architecture, called SimpleNet, based on a set of designing principles, with which we empirically show, a well-crafted yet simple and reasonably deep architecture can perform on par with deeper and more complex architectures. SimpleNet provides a good tradeoff between the computation/memory efficiency and the accuracy. Our simple 13-layer architecture outperforms most of the deeper and complex architectures to date such as VGGNet, ResNet, and GoogleNet on several well-known benchmarks while having 2 to 25 times fewer number of parameters and operations. This makes it very handy for embedded systems or systems with computational and memory limitations. We achieved state-of-the-art result on CIFAR10 outperforming several heavier architectures, near state of the art on MNIST and competitive results on CIFAR100 and SVHN. We also outperformed the much larger and deeper architectures such as VGGNet and popular variants of ResNets among others on the ImageNet dataset. Models are made available at: https://github.com/Coderx7/SimpleNet

Ali Rasekh, Erfan Bagheri Soula, Omid Daliran et al.

Despite significant advances in Multimodal Large Language Models (MLLMs), understanding complex temporal dynamics in videos remains a major challenge. Our experiments show that current Video Large Language Model (Video-LLM) architectures have critical limitations in temporal understanding, struggling with tasks that require detailed comprehension of action sequences and temporal progression. In this work, we propose a Video-LLM architecture that introduces stacked temporal attention modules directly within the vision encoder. This design incorporates a temporal attention in vision encoder, enabling the model to better capture the progression of actions and the relationships between frames before passing visual tokens to the LLM. Our results show that this approach significantly improves temporal reasoning and outperforms existing models in video question answering tasks, specifically in action recognition. We improve on benchmarks including VITATECS, MVBench, and Video-MME by up to +5.5%. By enhancing the vision encoder with temporal structure, we address a critical gap in video understanding for Video-LLMs. Project page and code are available at: https://alirasekh.github.io/STAVEQ2/.

Mohsen Fayyaz, Ali Modarressi, Hanieh Deilamsalehy et al.

Mixture-of-Experts (MoE) in Large Language Models (LLMs) routes each token through a subset of specialized Feed-Forward Networks (FFN), known as experts. We present SteerMoE, a framework for steering MoE models by detecting and controlling behavior-linked experts. Our detection method identifies experts with distinct activation patterns across paired inputs exhibiting contrasting behaviors. By selectively (de)activating such experts during inference, we control behaviors like faithfulness and safety without retraining or modifying weights. Across 11 benchmarks and 6 LLMs, our steering raises safety by up to +20% and faithfulness by +27%. In adversarial attack mode, it drops safety by -41% alone, and -100% when combined with existing jailbreak methods, bypassing all safety guardrails and exposing a new dimension of alignment faking hidden within experts.

Lucas Bandarkar, Chenyuan Yang, Mohsen Fayyaz et al.

Mixture-of-Experts (MoE) architectures have become the key to scaling modern LLMs, yet little is understood about how their sparse routing dynamics respond to multilingual data. In this work, we analyze expert routing patterns using parallel multilingual datasets and present highly interpretable layer-wise phenomena. We find that MoE models route tokens in language-specific ways in the early and late decoder layers but exhibit significant cross-lingual routing alignment in middle layers, mirroring parameter-sharing trends observed in dense LLMs. In particular, we reveal a clear, strong correlation between a model's performance in a given language and how similarly its tokens are routed to English in these layers. Extending beyond correlation, we explore inference-time interventions that induce higher cross-lingual routing alignment. We introduce a method that steers the router by promoting middle-layer task experts frequently activated in English, and it successfully increases multilingual performance. These 1-2% gains are remarkably consistent across two evaluation tasks, three models, and 15+ languages, especially given that these simple interventions override routers of extensively trained, state-of-the-art LLMs. In comparison, interventions outside of the middle layers or targeting multilingual-specialized experts only yield performance degradation. Altogether, we present numerous findings that explain how MoEs process non-English text and demonstrate that generalization is limited by the model's ability to leverage language-universal experts in all languages.

Liana Mikaelyan, Ayyoob Imani, Mathew Salvaris et al.

We introduce DeltaLLM, a new post-training compression technique to reduce the memory footprint of LLMs. We propose an alternative way of structuring LLMs with weight sharing between layers in subsequent Transformer blocks, along with additional low-rank difference matrices between them. For training, we adopt the progressing module replacement method and show that the lightweight training of the low-rank modules with approximately 30M-40M tokens is sufficient to achieve performance on par with LLMs of comparable sizes trained from scratch. We release the resultant models, DeltaLLAMA and DeltaPHI, with a 12% parameter reduction, retaining 90% of the performance of the base Llama and Phi models on common knowledge and reasoning benchmarks. Our method also outperforms compression techniques JointDrop, LaCo, ShortGPT and SliceGPT with the same number of parameters removed. For example, DeltaPhi 2.9B with a 24% reduction achieves similar average zero-shot accuracies as recovery fine-tuned SlicedPhi 3.3B with a 12% reduction, despite being approximately 400M parameters smaller with no fine-tuning applied. This work provides new insights into LLM architecture design and compression methods when storage space is critical.

Mohsen Fayyaz, Fan Yin, Jiao Sun et al.

We study how well large language models (LLMs) explain their generations through rationales -- a set of tokens extracted from the input text that reflect the decision-making process of LLMs. Specifically, we systematically study rationales derived using two approaches: (1) popular prompting-based methods, where prompts are used to guide LLMs in generating rationales, and (2) technical attribution-based methods, which leverage attention or gradients to identify important tokens. Our analysis spans three classification datasets with annotated rationales, encompassing tasks with varying performance levels. While prompting-based self-explanations are widely used, our study reveals that these explanations are not always as "aligned" with the human rationale as attribution-based explanations. Even more so, fine-tuning LLMs to enhance classification task accuracy does not enhance the alignment of prompting-based rationales. Still, it does considerably improve the alignment of attribution-based methods (e.g., InputXGradient). More importantly, we show that prompting-based self-explanation is also less "faithful" than attribution-based explanations, failing to provide a reliable account of the model's decision-making process. To evaluate faithfulness, unlike prior studies that excluded misclassified examples, we evaluate all instances and also examine the impact of fine-tuning and accuracy on alignment and faithfulness. Our findings suggest that inconclusive faithfulness results reported in earlier studies may stem from low classification accuracy. These findings underscore the importance of more rigorous and comprehensive evaluations of LLM rationales.

Ali Modarressi, Ayyoob Imani, Mohsen Fayyaz et al.

Large language models (LLMs) have significantly advanced the field of natural language processing (NLP) through their extensive parameters and comprehensive data utilization. However, existing LLMs lack a dedicated memory unit, limiting their ability to explicitly store and retrieve knowledge for various tasks. In this paper, we propose RET-LLM a novel framework that equips LLMs with a general write-read memory unit, allowing them to extract, store, and recall knowledge from the text as needed for task performance. Inspired by Davidsonian semantics theory, we extract and save knowledge in the form of triplets. The memory unit is designed to be scalable, aggregatable, updatable, and interpretable. Through qualitative evaluations, we demonstrate the superiority of our proposed framework over baseline approaches in question answering tasks. Moreover, our framework exhibits robust performance in handling temporal-based question answering tasks, showcasing its ability to effectively manage time-dependent information.

Mohsen Fayyaz, Soroush Abbasi Koohpayegani, Farnoush Rezaei Jafari et al.

While state-of-the-art vision transformer models achieve promising results in image classification, they are computationally expensive and require many GFLOPs. Although the GFLOPs of a vision transformer can be decreased by reducing the number of tokens in the network, there is no setting that is optimal for all input images. In this work, we therefore introduce a differentiable parameter-free Adaptive Token Sampler (ATS) module, which can be plugged into any existing vision transformer architecture. ATS empowers vision transformers by scoring and adaptively sampling significant tokens. As a result, the number of tokens is not constant anymore and varies for each input image. By integrating ATS as an additional layer within the current transformer blocks, we can convert them into much more efficient vision transformers with an adaptive number of tokens. Since ATS is a parameter-free module, it can be added to the off-the-shelf pre-trained vision transformers as a plug and play module, thus reducing their GFLOPs without any additional training. Moreover, due to its differentiable design, one can also train a vision transformer equipped with ATS. We evaluate the efficiency of our module in both image and video classification tasks by adding it to multiple SOTA vision transformers. Our proposed module improves the SOTA by reducing their computational costs (GFLOPs) by 2X, while preserving their accuracy on the ImageNet, Kinetics-400, and Kinetics-600 datasets.

Saber Pourheydari, Emad Bahrami, Mohsen Fayyaz et al.

While recurrent neural networks (RNNs) demonstrate outstanding capabilities for future video frame prediction, they model dynamics in a discrete time space, i.e., they predict the frames sequentially with a fixed temporal step. RNNs are therefore prone to accumulate the error as the number of future frames increases. In contrast, partial differential equations (PDEs) model physical phenomena like dynamics in a continuous time space. However, the estimated PDE for frame forecasting needs to be numerically solved, which is done by discretization of the PDE and diminishes most of the advantages compared to discrete models. In this work, we, therefore, propose to approximate the motion in a video by a continuous function using the Taylor series. To this end, we introduce TaylorSwiftNet, a novel convolutional neural network that learns to estimate the higher order terms of the Taylor series for a given input video. TaylorSwiftNet can swiftly predict future frames in parallel and it allows to change the temporal resolution of the forecast frames on-the-fly. The experimental results on various datasets demonstrate the superiority of our model.

Nadine Behrmann, Mohsen Fayyaz, Juergen Gall et al.

Self-supervised video representation methods typically focus on the representation of temporal attributes in videos. However, the role of stationary versus non-stationary attributes is less explored: Stationary features, which remain similar throughout the video, enable the prediction of video-level action classes. Non-stationary features, which represent temporally varying attributes, are more beneficial for downstream tasks involving more fine-grained temporal understanding, such as action segmentation. We argue that a single representation to capture both types of features is sub-optimal, and propose to decompose the representation space into stationary and non-stationary features via contrastive learning from long and short views, i.e. long video sequences and their shorter sub-sequences. Stationary features are shared between the short and long views, while non-stationary features aggregate the short views to match the corresponding long view. To empirically verify our approach, we demonstrate that our stationary features work particularly well on an action recognition downstream task, while our non-stationary features perform better on action segmentation. Furthermore, we analyse the learned representations and find that stationary features capture more temporally stable, static attributes, while non-stationary features encompass more temporally varying ones.

Mohsen Fayyaz, Ehsan Aghazadeh, Ali Modarressi et al.

Most of the recent works on probing representations have focused on BERT, with the presumption that the findings might be similar to the other models. In this work, we extend the probing studies to two other models in the family, namely ELECTRA and XLNet, showing that variations in the pre-training objectives or architectural choices can result in different behaviors in encoding linguistic information in the representations. Most notably, we observe that ELECTRA tends to encode linguistic knowledge in the deeper layers, whereas XLNet instead concentrates that in the earlier layers. Also, the former model undergoes a slight change during fine-tuning, whereas the latter experiences significant adjustments. Moreover, we show that drawing conclusions based on the weight mixing evaluation strategy -- which is widely used in the context of layer-wise probing -- can be misleading given the norm disparity of the representations across different layers. Instead, we adopt an alternative information-theoretic probing with minimum description length, which has recently been proven to provide more reliable and informative results.

Mohsen Fayyaz, Emad Bahrami, Ali Diba et al.

While state-of-the-art 3D Convolutional Neural Networks (CNN) achieve very good results on action recognition datasets, they are computationally very expensive and require many GFLOPs. While the GFLOPs of a 3D CNN can be decreased by reducing the temporal feature resolution within the network, there is no setting that is optimal for all input clips. In this work, we therefore introduce a differentiable Similarity Guided Sampling (SGS) module, which can be plugged into any existing 3D CNN architecture. SGS empowers 3D CNNs by learning the similarity of temporal features and grouping similar features together. As a result, the temporal feature resolution is not anymore static but it varies for each input video clip. By integrating SGS as an additional layer within current 3D CNNs, we can convert them into much more efficient 3D CNNs with adaptive temporal feature resolutions (ATFR). Our evaluations show that the proposed module improves the state-of-the-art by reducing the computational cost (GFLOPs) by half while preserving or even improving the accuracy. We evaluate our module by adding it to multiple state-of-the-art 3D CNNs on various datasets such as Kinetics-600, Kinetics-400, mini-Kinetics, Something-Something V2, UCF101, and HMDB51.

Mohsen Fayyaz, Juergen Gall

Temporal action segmentation is a topic of increasing interest, however, annotating each frame in a video is cumbersome and costly. Weakly supervised approaches therefore aim at learning temporal action segmentation from videos that are only weakly labeled. In this work, we assume that for each training video only the list of actions is given that occur in the video, but not when, how often, and in which order they occur. In order to address this task, we propose an approach that can be trained end-to-end on such data. The approach divides the video into smaller temporal regions and predicts for each region the action label and its length. In addition, the network estimates the action labels for each frame. By measuring how consistent the frame-wise predictions are with respect to the temporal regions and the annotated action labels, the network learns to divide a video into class-consistent regions. We evaluate our approach on three datasets where the approach achieves state-of-the-art results.

Ali Diba, Mohsen Fayyaz, Vivek Sharma et al.

Video recognition has been advanced in recent years by benchmarks with rich annotations. However, research is still mainly limited to human action or sports recognition - focusing on a highly specific video understanding task and thus leaving a significant gap towards describing the overall content of a video. We fill this gap by presenting a large-scale "Holistic Video Understanding Dataset"~(HVU). HVU is organized hierarchically in a semantic taxonomy that focuses on multi-label and multi-task video understanding as a comprehensive problem that encompasses the recognition of multiple semantic aspects in the dynamic scene. HVU contains approx.~572k videos in total with 9 million annotations for training, validation, and test set spanning over 3142 labels. HVU encompasses semantic aspects defined on categories of scenes, objects, actions, events, attributes, and concepts which naturally captures the real-world scenarios. We demonstrate the generalization capability of HVU on three challenging tasks: 1.) Video classification, 2.) Video captioning and 3.) Video clustering tasks. In particular for video classification, we introduce a new spatio-temporal deep neural network architecture called "Holistic Appearance and Temporal Network"~(HATNet) that builds on fusing 2D and 3D architectures into one by combining intermediate representations of appearance and temporal cues. HATNet focuses on the multi-label and multi-task learning problem and is trained in an end-to-end manner. Via our experiments, we validate the idea that holistic representation learning is complementary, and can play a key role in enabling many real-world applications.

Yaser Souri, Mohsen Fayyaz, Luca Minciullo et al.

Action segmentation is the task of predicting the actions for each frame of a video. As obtaining the full annotation of videos for action segmentation is expensive, weakly supervised approaches that can learn only from transcripts are appealing. In this paper, we propose a novel end-to-end approach for weakly supervised action segmentation based on a two-branch neural network. The two branches of our network predict two redundant but different representations for action segmentation and we propose a novel mutual consistency (MuCon) loss that enforces the consistency of the two redundant representations. Using the MuCon loss together with a loss for transcript prediction, our proposed approach achieves the accuracy of state-of-the-art approaches while being $14$ times faster to train and $20$ times faster during inference. The MuCon loss proves beneficial even in the fully supervised setting.

Mohammad Hajizadeh Saffar, Mohsen Fayyaz, Mohammad Sabokrou et al.

This paper presents an accurate method for verifying online signatures. The main difficulty of signature verification come from: (1) Lacking enough training samples (2) The methods must be spatial change invariant. To deal with these difficulties and modeling the signatures efficiently, we propose a method that a one-class classifier per each user is built on discriminative features. First, we pre-train a sparse auto-encoder using a large number of unlabeled signatures, then we applied the discriminative features, which are learned by auto-encoder to represent the training and testing signatures as a self-thought learning method (i.e. we have introduced a signature descriptor). Finally, user's signatures are modeled and classified using a one-class classifier. The proposed method is independent on signature datasets thanks to self-taught learning. The experimental results indicate significant error reduction and accuracy enhancement in comparison with state-of-the-art methods on SVC2004 and SUSIG datasets.

Ali Diba, Mohsen Fayyaz, Vivek Sharma et al.

The work in this paper is driven by the question if spatio-temporal correlations are enough for 3D convolutional neural networks (CNN)? Most of the traditional 3D networks use local spatio-temporal features. We introduce a new block that models correlations between channels of a 3D CNN with respect to temporal and spatial features. This new block can be added as a residual unit to different parts of 3D CNNs. We name our novel block 'Spatio-Temporal Channel Correlation' (STC). By embedding this block to the current state-of-the-art architectures such as ResNext and ResNet, we improved the performance by 2-3\% on Kinetics dataset. Our experiments show that adding STC blocks to current state-of-the-art architectures outperforms the state-of-the-art methods on the HMDB51, UCF101 and Kinetics datasets. The other issue in training 3D CNNs is about training them from scratch with a huge labeled dataset to get a reasonable performance. So the knowledge learned in 2D CNNs is completely ignored. Another contribution in this work is a simple and effective technique to transfer knowledge from a pre-trained 2D CNN to a randomly initialized 3D CNN for a stable weight initialization. This allows us to significantly reduce the number of training samples for 3D CNNs. Thus, by fine-tuning this network, we beat the performance of generic and recent methods in 3D CNNs, which were trained on large video datasets, e.g. Sports-1M, and fine-tuned on the target datasets, e.g. HMDB51/UCF101.

Mohammad Hajizadeh Saffar, Mohsen Fayyaz, Mohammad Sabokrou et al.

This paper gives an overview on semantic segmentation consists of an explanation of this field, it's status and relation with other vision fundamental tasks, different datasets and common evaluation parameters that have been used by researchers. This survey also includes an overall review on a variety of recent approaches (RDF, MRF, CRF, etc.) and their advantages and challenges and shows the superiority of CNN-based semantic segmentation systems on CamVid and NYUDv2 datasets. In addition, some areas that is ideal for future work have mentioned.

Mohammad Sabokrou, Masoud Pourreza, Mohsen Fayyaz et al.

Real-time detection of irregularities in visual data is very invaluable and useful in many prospective applications including surveillance, patient monitoring systems, etc. With the surge of deep learning methods in the recent years, researchers have tried a wide spectrum of methods for different applications. However, for the case of irregularity or anomaly detection in videos, training an end-to-end model is still an open challenge, since often irregularity is not well-defined and there are not enough irregular samples to use during training. In this paper, inspired by the success of generative adversarial networks (GANs) for training deep models in unsupervised or self-supervised settings, we propose an end-to-end deep network for detection and fine localization of irregularities in videos (and images). Our proposed architecture is composed of two networks, which are trained in competing with each other while collaborating to find the irregularity. One network works as a pixel-level irregularity Inpainter, and the other works as a patch-level Detector. After an adversarial self-supervised training, in which I tries to fool D into accepting its inpainted output as regular (normal), the two networks collaborate to detect and fine-segment the irregularity in any given testing video. Our results on three different datasets show that our method can outperform the state-of-the-art and fine-segment the irregularity.

Ali Diba, Mohsen Fayyaz, Vivek Sharma et al.

The work in this paper is driven by the question how to exploit the temporal cues available in videos for their accurate classification, and for human action recognition in particular? Thus far, the vision community has focused on spatio-temporal approaches with fixed temporal convolution kernel depths. We introduce a new temporal layer that models variable temporal convolution kernel depths. We embed this new temporal layer in our proposed 3D CNN. We extend the DenseNet architecture - which normally is 2D - with 3D filters and pooling kernels. We name our proposed video convolutional network `Temporal 3D ConvNet'~(T3D) and its new temporal layer `Temporal Transition Layer'~(TTL). Our experiments show that T3D outperforms the current state-of-the-art methods on the HMDB51, UCF101 and Kinetics datasets. The other issue in training 3D ConvNets is about training them from scratch with a huge labeled dataset to get a reasonable performance. So the knowledge learned in 2D ConvNets is completely ignored. Another contribution in this work is a simple and effective technique to transfer knowledge from a pre-trained 2D CNN to a randomly initialized 3D CNN for a stable weight initialization. This allows us to significantly reduce the number of training samples for 3D CNNs. Thus, by finetuning this network, we beat the performance of generic and recent methods in 3D CNNs, which were trained on large video datasets, e.g. Sports-1M, and finetuned on the target datasets, e.g. HMDB51/UCF101. The T3D codes will be released

Mohammad Sabokrou, Mohsen Fayyaz, Mahmood Fathy et al.

The detection of abnormal behaviours in crowded scenes has to deal with many challenges. This paper presents an efficient method for detection and localization of anomalies in videos. Using fully convolutional neural networks (FCNs) and temporal data, a pre-trained supervised FCN is transferred into an unsupervised FCN ensuring the detection of (global) anomalies in scenes. High performance in terms of speed and accuracy is achieved by investigating the cascaded detection as a result of reducing computation complexities. This FCN-based architecture addresses two main tasks, feature representation and cascaded outlier detection. Experimental results on two benchmarks suggest that detection and localization of the proposed method outperforms existing methods in terms of accuracy.

Mohsen Fayyaz, Mohammad Hajizadeh Saffar, Mohammad Sabokrou et al.

This paper presents a novel method to involve both spatial and temporal features for semantic video segmentation. Current work on convolutional neural networks(CNNs) has shown that CNNs provide advanced spatial features supporting a very good performance of solutions for both image and video analysis, especially for the semantic segmentation task. We investigate how involving temporal features also has a good effect on segmenting video data. We propose a module based on a long short-term memory (LSTM) architecture of a recurrent neural network for interpreting the temporal characteristics of video frames over time. Our system takes as input frames of a video and produces a correspondingly-sized output; for segmenting the video our method combines the use of three components: First, the regional spatial features of frames are extracted using a CNN; then, using LSTM the temporal features are added; finally, by deconvolving the spatio-temporal features we produce pixel-wise predictions. Our key insight is to build spatio-temporal convolutional networks (spatio-temporal CNNs) that have an end-to-end architecture for semantic video segmentation. We adapted fully some known convolutional network architectures (such as FCN-AlexNet and FCN-VGG16), and dilated convolution into our spatio-temporal CNNs. Our spatio-temporal CNNs achieve state-of-the-art semantic segmentation, as demonstrated for the Camvid and NYUDv2 datasets.

Mohsen Fayyaz, Masoud PourReza, Mohammad Hajizadeh Saffar et al.

In this paper, we propose a method for user Finger Vein Authentication (FVA) as a biometric system. Using the discriminative features for classifying theses finger veins is one of the main tips that make difference in related works, Thus we propose to learn a set of representative features, based on autoencoders. We model the user finger vein using a Gaussian distribution. Experimental results show that our algorithm perform like a state-of-the-art on SDUMLA-HMT benchmark.

Mohsen Fayyaz, Mohammad Hajizadeh_Saffar, Mohammad Sabokrou et al.

Biometrics systems have been used in a wide range of applications and have improved people authentication. Signature verification is one of the most common biometric methods with techniques that employ various specifications of a signature. Recently, deep learning has achieved great success in many fields, such as image, sounds and text processing. In this paper, deep learning method has been used for feature extraction and feature selection.