Alexandros Iosifidis

Firas Laakom, Jenni Raitoharju, Alexandros Iosifidis et al.

Neural networks are composed of multiple layers arranged in a hierarchical structure jointly trained with a gradient-based optimization, where the errors are back-propagated from the last layer back to the first one. At each optimization step, neurons at a given layer receive feedback from neurons belonging to higher layers of the hierarchy. In this paper, we propose to complement this traditional 'between-layer' feedback with additional 'within-layer' feedback to encourage the diversity of the activations within the same layer. To this end, we measure the pairwise similarity between the outputs of the neurons and use it to model the layer's overall diversity. We present an extensive empirical study confirming that the proposed approach enhances the performance of several state-of-the-art neural network models in multiple tasks. The code is publically available at \url{https://github.com/firasl/AAAI-23-WLD-Reg}

Lukas Hedegaard, Aman Alok, Juby Jose et al.

Adapters are a parameter-efficient alternative to fine-tuning, which augment a frozen base network to learn new tasks. Yet, the inference of the adapted model is often slower than the corresponding fine-tuned model. To improve on this, we propose Structured Pruning Adapters (SPAs), a family of compressing, task-switching network adapters, that accelerate and specialize networks using tiny parameter sets and structured pruning. Specifically, we propose a channel-based SPA and evaluate it with a suite of pruning methods on multiple computer vision benchmarks. Compared to regular structured pruning with fine-tuning, our channel-SPAs improve accuracy by 6.9% on average while using half the parameters at 90% pruned weights. Alternatively, they can learn adaptations with 17x fewer parameters at 70% pruning with 1.6% lower accuracy. Similarly, our block-SPA requires far fewer parameters than pruning with fine-tuning. Our experimental code and Python library of adapters are available at github.com/lukashedegaard/structured-pruning-adapters.

Lukas Hedegaard, Alexandros Iosifidis

We present Continual Inference, a Python library for implementing Continual Inference Networks (CINs) in PyTorch, a class of Neural Networks designed specifically for efficient inference in both online and batch processing scenarios. We offer a comprehensive introduction and guide to CINs and their implementation in practice, and provide best-practices and code examples for composing complex modules for modern Deep Learning. Continual Inference is readily downloadable via the Python Package Index and at \url{www.github.com/lukashedegaard/continual-inference}.

Nan Li, Alexandros Iosifidis, Qi Zhang

This paper studies the computational offloading of CNN inference in device-edge co-inference systems. Inspired by the emerging paradigm semantic communication, we propose a novel autoencoder-based CNN architecture (AECNN), for effective feature extraction at end-device. We design a feature compression module based on the channel attention method in CNN, to compress the intermediate data by selecting the most important features. To further reduce communication overhead, we can use entropy encoding to remove the statistical redundancy in the compressed data. At the receiver, we design a lightweight decoder to reconstruct the intermediate data through learning from the received compressed data to improve accuracy. To fasten the convergence, we use a step-by-step approach to train the neural networks obtained based on ResNet-50 architecture. Experimental results show that AECNN can compress the intermediate data by more than 256x with only about 4% accuracy loss, which outperforms the state-of-the-art work, BottleNet++. Compared to offloading inference task directly to edge server, AECNN can complete inference task earlier, in particular, under poor wireless channel condition, which highlights the effectiveness of AECNN in guaranteeing higher accuracy within time constraint.

Nan Li, Alexandros Iosifidis, Qi Zhang

This paper studies the computational offloading of CNN inference in dynamic multi-access edge computing (MEC) networks. To address the uncertainties in communication time and Edge servers' available capacity, we use early-exit mechanism to terminate the computation earlier to meet the deadline of inference tasks. We design a reward function to trade off the communication, computation and inference accuracy, and formulate the offloading problem of CNN inference as a maximization problem with the goal of maximizing the average inference accuracy and throughput in long term. To solve the maximization problem, we propose a graph reinforcement learning-based early-exit mechanism (GRLE), which outperforms the state-of-the-art work, deep reinforcement learning-based online offloading (DROO) and its enhanced method, DROO with early-exit mechanism (DROOE), under different dynamic scenarios. The experimental results show that GRLE achieves the average accuracy up to 3.41x over graph reinforcement learning (GRL) and 1.45x over DROOE, which shows the advantages of GRLE for offloading decision-making in dynamic MEC.

Nan Li, Alexandros Iosifidis, Qi Zhang

This paper studies inference acceleration using distributed convolutional neural networks (CNNs) in collaborative edge computing. To ensure inference accuracy in inference task partitioning, we consider the receptive-field when performing segment-based partitioning. To maximize the parallelization between the communication and computing processes, thereby minimizing the total inference time of an inference task, we design a novel task collaboration scheme in which the overlapping zone of the sub-tasks on secondary edge servers (ESs) is executed on the host ES, named as HALP. We further extend HALP to the scenario of multiple tasks. Experimental results show that HALP can accelerate CNN inference in VGG-16 by 1.7-2.0x for a single task and 1.7-1.8x for 4 tasks per batch on GTX 1080TI and JETSON AGX Xavier, which outperforms the state-of-the-art work MoDNN. Moreover, we evaluate the service reliability under time-variant channel, which shows that HALP is an effective solution to ensure high service reliability with strict service deadline.

Martin Magris, Alexandros Iosifidis

The last decade witnessed a growing interest in Bayesian learning. Yet, the technicality of the topic and the multitude of ingredients involved therein, besides the complexity of turning theory into practical implementations, limit the use of the Bayesian learning paradigm, preventing its widespread adoption across different fields and applications. This self-contained survey engages and introduces readers to the principles and algorithms of Bayesian Learning for Neural Networks. It provides an introduction to the topic from an accessible, practical-algorithmic perspective. Upon providing a general introduction to Bayesian Neural Networks, we discuss and present both standard and recent approaches for Bayesian inference, with an emphasis on solutions relying on Variational Inference and the use of Natural gradients. We also discuss the use of manifold optimization as a state-of-the-art approach to Bayesian learning. We examine the characteristic properties of all the discussed methods, and provide pseudo-codes for their implementation, paying attention to practical aspects, such as the computation of the gradients.

Mirgita Frasheri, Arian Bakhtiarnia, Lukas Esterle et al.

Countless terms of service (ToS) are being signed everyday by users all over the world while interacting with all kinds of apps and websites. More often than not, these online contracts spanning double-digit pages are signed blindly by users who simply want immediate access to the desired service. What would normally require a consultation with a legal team, has now become a mundane activity consisting of a few clicks where users potentially sign away their rights, for instance in terms of their data privacy, to countless online entities/companies. Large language models (LLMs) are good at parsing long text-based documents, and could potentially be adopted to help users when dealing with dubious clauses in ToS and their underlying privacy policies. To investigate the utility of existing models for this task, we first build a dataset consisting of 12 questions applied individually to a set of privacy policies crawled from popular websites. Thereafter, a series of open-source as well as commercial chatbots such as ChatGPT, are queried over each question, with the answers being compared to a given ground truth. Our results show that some open-source models are able to provide a higher accuracy compared to some commercial models. However, the best performance is recorded from a commercial chatbot (ChatGPT4). Overall, all models perform only slightly better than random at this task. Consequently, their performance needs to be significantly improved before they can be adopted at large for this purpose.

Mehdi Rafiei, Toby P. Breckon, Alexandros Iosifidis

Anomaly detection methods have demonstrated remarkable success across various applications. However, assessing their performance, particularly at the pixel-level, presents a complex challenge due to the severe imbalance that is most commonly present between normal and abnormal samples. Commonly adopted evaluation metrics designed for pixel-level detection may not effectively capture the nuanced performance variations arising from this class imbalance. In this paper, we dissect the intricacies of this challenge, underscored by visual evidence and statistical analysis, leading to delve into the need for evaluation metrics that account for the imbalance. We offer insights into more accurate metrics, using eleven leading contemporary anomaly detection methods on twenty-one anomaly detection problems. Overall, from this extensive experimental evaluation, we can conclude that Precision-Recall-based metrics can better capture relative method performance, making them more suitable for the task.

Firas Laakom, Jenni Raitoharju, Alexandros Iosifidis et al.

Despite the superior performance of CNN, deploying them on low computational power devices is still limited as they are typically computationally expensive. One key cause of the high complexity is the connection between the convolution layers and the fully connected layers, which typically requires a high number of parameters. To alleviate this issue, Bag of Features (BoF) pooling has been recently proposed. BoF learns a dictionary, that is used to compile a histogram representation of the input. In this paper, we propose an approach that builds on top of BoF pooling to boost its efficiency by ensuring that the items of the learned dictionary are non-redundant. We propose an additional loss term, based on the pair-wise correlation of the items of the dictionary, which complements the standard loss to explicitly regularize the model to learn a more diverse and rich dictionary. The proposed strategy yields an efficient variant of BoF and further boosts its performance, without any additional parameters.

Illia Oleksiienko, Paraskevi Nousi, Nikolaos Passalis et al.

In this paper, we propose a novel voxel-based 3D single object tracking (3D SOT) method called Voxel Pseudo Image Tracking (VPIT). VPIT is the first method that uses voxel pseudo images for 3D SOT. The input point cloud is structured by pillar-based voxelization, and the resulting pseudo image is used as an input to a 2D-like Siamese SOT method. The pseudo image is created in the Bird's-eye View (BEV) coordinates, and therefore the objects in it have constant size. Thus, only the object rotation can change in the new coordinate system and not the object scale. For this reason, we replace multi-scale search with a multi-rotation search, where differently rotated search regions are compared against a single target representation to predict both position and rotation of the object. Experiments on KITTI Tracking dataset show that VPIT is the fastest 3D SOT method and maintains competitive Success and Precision values. Application of a SOT method in a real-world scenario meets with limitations such as lower computational capabilities of embedded devices and a latency-unforgiving environment, where the method is forced to skip certain data frames if the inference speed is not high enough. We implement a real-time evaluation protocol and show that other methods lose most of their performance on embedded devices, while VPIT maintains its ability to track the object.

Firas Laakom, Jenni Raitoharju, Alexandros Iosifidis et al.

Energy-based learning is a powerful learning paradigm that encapsulates various discriminative and generative approaches. An energy-based model (EBM) is typically formed of inner-model(s) that learn a combination of the different features to generate an energy mapping for each input configuration. In this paper, we focus on the diversity of the produced feature set. We extend the probably approximately correct (PAC) theory of EBMs and analyze the effect of redundancy reduction on the performance of EBMs. We derive generalization bounds for various learning contexts, i.e., regression, classification, and implicit regression, with different energy functions and we show that indeed reducing redundancy of the feature set can consistently decrease the gap between the true and empirical expectation of the energy and boosts the performance of the model.

Andrea Cavagna, Nan Li, Alexandros Iosifidis et al.

This paper aims to design robust Edge Intelligence using semantic communication for time-critical IoT applications. We systematically analyze the effect of image DCT coefficients on inference accuracy and propose the channel-agnostic effectiveness encoding for offloading by transmitting the most meaningful task data first. This scheme can well utilize all available communication resource and strike a balance between transmission latency and inference accuracy. Then, we design an effectiveness decoding by implementing a novel image augmentation process for convolutional neural network (CNN) training, through which an original CNN model is transformed into a Robust CNN model. We use the proposed training method to generate Robust MobileNet-v2 and Robust ResNet-50. The proposed Edge Intelligence framework consists of the proposed effectiveness encoding and effectiveness decoding. The experimental results show that the effectiveness decoding using the Robust CNN models perform consistently better under various image distortions caused by channel errors or limited communication resource. The proposed Edge Intelligence framework using semantic communication significantly outperforms the conventional approach under latency and data rate constraints, in particular, under ultra stringent deadlines and low data rate.

Nan Li, Alexandros Iosifidis, Qi Zhang

This paper studies inference acceleration using distributed convolutional neural networks (CNNs) in collaborative edge computing network. To avoid inference accuracy loss in inference task partitioning, we propose receptive field-based segmentation (RFS). To reduce the computation time and communication overhead, we propose a novel collaborative edge computing using fused-layer parallelization to partition a CNN model into multiple blocks of convolutional layers. In this scheme, the collaborative edge servers (ESs) only need to exchange small fraction of the sub-outputs after computing each fused block. In addition, to find the optimal solution of partitioning a CNN model into multiple blocks, we use dynamic programming, named as dynamic programming for fused-layer parallelization (DPFP). The experimental results show that DPFP can accelerate inference of VGG-16 up to 73% compared with the pre-trained model, which outperforms the existing work MoDNN in all tested scenarios. Moreover, we evaluate the service reliability of DPFP under time-variant channel, which shows that DPFP is an effective solution to ensure high service reliability with strict service deadline.

Arian Bakhtiarnia, Nemanja Milošević, Qi Zhang et al.

Deploying deep neural networks (DNNs) on IoT and mobile devices is a challenging task due to their limited computational resources. Thus, demanding tasks are often entirely offloaded to edge servers which can accelerate inference, however, it also causes communication cost and evokes privacy concerns. In addition, this approach leaves the computational capacity of end devices unused. Split computing is a paradigm where a DNN is split into two sections; the first section is executed on the end device, and the output is transmitted to the edge server where the final section is executed. Here, we introduce dynamic split computing, where the optimal split location is dynamically selected based on the state of the communication channel. By using natural bottlenecks that already exist in modern DNN architectures, dynamic split computing avoids retraining and hyperparameter optimization, and does not have any negative impact on the final accuracy of DNNs. Through extensive experiments, we show that dynamic split computing achieves faster inference in edge computing environments where the data rate and server load vary over time.

Arian Bakhtiarnia, Qi Zhang, Alexandros Iosifidis

Cameras in modern devices such as smartphones, satellites and medical equipment are capable of capturing very high resolution images and videos. Such high-resolution data often need to be processed by deep learning models for cancer detection, automated road navigation, weather prediction, surveillance, optimizing agricultural processes and many other applications. Using high-resolution images and videos as direct inputs for deep learning models creates many challenges due to their high number of parameters, computation cost, inference latency and GPU memory consumption. Simple approaches such as resizing the images to a lower resolution are common in the literature, however, they typically significantly decrease accuracy. Several works in the literature propose better alternatives in order to deal with the challenges of high-resolution data and improve accuracy and speed while complying with hardware limitations and time restrictions. This survey describes such efficient high-resolution deep learning methods, summarizes real-world applications of high-resolution deep learning, and provides comprehensive information about available high-resolution datasets.

Lukas Hedegaard, Negar Heidari, Alexandros Iosifidis

Graph-based reasoning over skeleton data has emerged as a promising approach for human action recognition. However, the application of prior graph-based methods, which predominantly employ whole temporal sequences as their input, to the setting of online inference entails considerable computational redundancy. In this paper, we tackle this issue by reformulating the Spatio-Temporal Graph Convolutional Neural Network as a Continual Inference Network, which can perform step-by-step predictions in time without repeat frame processing. To evaluate our method, we create a continual version of ST-GCN, CoST-GCN, alongside two derived methods with different self-attention mechanisms, CoAGCN and CoS-TR. We investigate weight transfer strategies and architectural modifications for inference acceleration, and perform experiments on the NTU RGB+D 60, NTU RGB+D 120, and Kinetics Skeleton 400 datasets. Retaining similar predictive accuracy, we observe up to 109x reduction in time complexity, on-hardware accelerations of 26x, and reductions in maximum allocated memory of 52% during online inference.

Mehdi Rafiei, Jenni Raitoharju, Alexandros Iosifidis

X-ray imaging technology has been used for decades in clinical tasks to reveal the internal condition of different organs, and in recent years, it has become more common in other areas such as industry, security, and geography. The recent development of computer vision and machine learning techniques has also made it easier to automatically process X-ray images and several machine learning-based object (anomaly) detection, classification, and segmentation methods have been recently employed in X-ray image analysis. Due to the high potential of deep learning in related image processing applications, it has been used in most of the studies. This survey reviews the recent research on using computer vision and machine learning for X-ray analysis in industrial production and security applications and covers the applications, techniques, evaluation metrics, datasets, and performance comparison of those techniques on publicly available datasets. We also highlight some drawbacks in the published research and give recommendations for future research in computer vision-based X-ray analysis.

Zhongtian Dong, Nan Li, Alexandros Iosifidis et al.

For time-critical IoT applications using deep learning, inference acceleration through distributed computing is a promising approach to meet a stringent deadline. In this paper, we implement a working prototype of a new distributed inference acceleration method HALP using three raspberry Pi 4. HALP accelerates inference by designing a seamless collaboration among edge devices (EDs) in Edge Computing. We maximize the parallelization between communication and computation among the collaborative EDs by optimizing the task partitioning ratio based on the segment-based partitioning. Experimental results show that the distributed inference HALP achieves 1.7x inference acceleration for VGG-16. Then, we combine distributed inference with conventional neural network model compression by setting up different shrinking hyperparameters for MobileNet-V1. In this way, we can further accelerate inference but at the cost of inference accuracy loss. To strike a balance between latency and accuracy, we propose dynamic model selection to select a model which provides the highest accuracy within the latency constraint. It is shown that the model selection with distributed inference HALP can significantly improve service reliability compared to the conventional stand-alone computation.

Mehdi Rafiei, Dat Thanh Tran, Alexandros Iosifidis

Mineral wool production is a non-linear process that makes it hard to control the final quality. Therefore, having a non-destructive method to analyze the product quality and recognize defective products is critical. For this purpose, we developed a visual quality control system for mineral wool. X-ray images of wool specimens were collected to create a training set of defective and non-defective samples. Afterward, we developed several recognition models based on the ResNet architecture to find the most efficient model. In order to have a light-weight and fast inference model for real-life applicability, two structural pruning methods are applied to the classifiers. Considering the low quantity of the dataset, cross-validation and augmentation methods are used during the training. As a result, we obtained a model with more than 98% accuracy, which in comparison to the current procedure used at the company, it can recognize 20% more defective products.

Negar Heidari, Alexandros Iosifidis

Diversity of the features extracted by deep neural networks is important for enhancing the model generalization ability and accordingly its performance in different learning tasks. Facial expression recognition in the wild has attracted interest in recent years due to the challenges existing in this area for extracting discriminative and informative features from occluded images in real-world scenarios. In this paper, we propose a mechanism to diversify the features extracted by CNN layers of state-of-the-art facial expression recognition architectures for enhancing the model capacity in learning discriminative features. To evaluate the effectiveness of the proposed approach, we incorporate this mechanism in two state-of-the-art models to (i) diversify local/global features in an attention-based model and (ii) diversify features extracted by different learners in an ensemble-based model. Experimental results on three well-known facial expression recognition in-the-wild datasets, AffectNet, FER+, and RAF-DB, show the effectiveness of our method, achieving the state-of-the-art performance of 89.99% on RAF-DB, 89.34% on FER+ and the competitive accuracy of 60.02% on AffectNet dataset.

Elizabeth Fons, Alejandro Sztrajman, Yousef El-laham et al.

Implicit neural representations (INRs) have recently emerged as a powerful tool that provides an accurate and resolution-independent encoding of data. Their robustness as general approximators has been shown in a wide variety of data sources, with applications on image, sound, and 3D scene representation. However, little attention has been given to leveraging these architectures for the representation and analysis of time series data. In this paper, we analyze the representation of time series using INRs, comparing different activation functions in terms of reconstruction accuracy and training convergence speed. We show how these networks can be leveraged for the imputation of time series, with applications on both univariate and multivariate data. Finally, we propose a hypernetwork architecture that leverages INRs to learn a compressed latent representation of an entire time series dataset. We introduce an FFT-based loss to guide training so that all frequencies are preserved in the time series. We show that this network can be used to encode time series as INRs, and their embeddings can be interpolated to generate new time series from existing ones. We evaluate our generative method by using it for data augmentation, and show that it is competitive against current state-of-the-art approaches for augmentation of time series.

Firas Laakom, Fahad Sohrab, Jenni Raitoharju et al.

One-class classification refers to approaches of learning using data from a single class only. In this paper, we propose a deep learning one-class classification method suitable for multimodal data, which relies on two convolutional autoencoders jointly trained to reconstruct the positive input data while obtaining the data representations in the latent space as compact as possible. During inference, the distance of the latent representation of an input to the origin can be used as an anomaly score. Experimental results using a multimodal macroinvertebrate image classification dataset show that the proposed multimodal method yields better results as compared to the unimodal approach. Furthermore, study the effect of different input image sizes, and we investigate how recently proposed feature diversity regularizers affect the performance of our approach. We show that such regularizers improve performance.

Jenni Raitoharju, Alexandros Iosifidis

In this paper, we formulate a new generalized reference kernel hoping to improve the original base kernel using a set of reference vectors. Depending on the selected reference vectors, our formulation shows similarities to approximate kernels, random mappings, and Non-linear Projection Trick. Focusing on small-scale one-class classification, our analysis and experimental results show that the new formulation provides approaches to regularize, adjust the rank, and incorporate additional information into the kernel itself, leading to improved one-class classification accuracy.

Illia Oleksiienko, Paraskevi Nousi, Nikolaos Passalis et al.

Uncertainty estimation is an important task for critical problems, such as robotics and autonomous driving, because it allows creating statistically better perception models and signaling the model's certainty in its predictions to the decision method or a human supervisor. In this paper, we propose a Variational Neural Network-based version of a Voxel Pseudo Image Tracking (VPIT) method for 3D Single Object Tracking. The Variational Feature Generation Network of the proposed Variational VPIT computes features for target and search regions and the corresponding uncertainties, which are later combined using an uncertainty-aware cross-correlation module in one of two ways: by computing similarity between the corresponding uncertainties and adding it to the regular cross-correlation values, or by penalizing the uncertain feature channels to increase influence of the certain features. In experiments, we show that both methods improve tracking performance, while penalization of uncertain features provides the best uncertainty quality.

Muhammad Azlan Qazi, Alexandros Iosifidis, Qi Zhang

Distributing Transformer inference across embedded edge devices can alleviate individual memory and compute constraints, yet practical benefits on real hardware remain unclear: prior work relies largely on simulations that overlook hardware-specific communication overheads. We present a hardware prototype study on NVIDIA Jetson Orin Nano devices connected over WiFi. Our key finding is that the dominant bottleneck is not just network bandwidth but also the CPU-GPU staging during communication. Because Jetson's integrated GPU architecture lacks the PCIe/NVLink pathway that NCCL requires, all inter-device data communication should be routed through GLOO and staged in CPU memory; an overhead that scales with communication data volume and makes full-tensor exchange slower than single-device inference across the batch sizes for medium sized models such as ViT. We therefore evaluate Prism by combining Segment Means compression with lightweight offline profiling to adaptively select between local and distributed execution at runtime. Experiments show that this strategy reduces latency by 65%-77% and energy consumption by 34%-52% relative to full-tensor exchange in static distributed execution setup, demonstrating that profiling-driven adaptation is essential for practical distributed Transformer inference on embedded hardware.

Mostafa Shabani, Dat Thanh Tran, Juho Kanniainen et al.

Deep Learning models have become dominant in tackling financial time-series analysis problems, overturning conventional machine learning and statistical methods. Most often, a model trained for one market or security cannot be directly applied to another market or security due to differences inherent in the market conditions. In addition, as the market evolves through time, it is necessary to update the existing models or train new ones when new data is made available. This scenario, which is inherent in most financial forecasting applications, naturally raises the following research question: How to efficiently adapt a pre-trained model to a new set of data while retaining performance on the old data, especially when the old data is not accessible? In this paper, we propose a method to efficiently retain the knowledge available in a neural network pre-trained on a set of securities and adapt it to achieve high performance in new ones. In our method, the prior knowledge encoded in a pre-trained neural network is maintained by keeping existing connections fixed, and this knowledge is adjusted for the new securities by a set of augmented connections, which are optimized using the new data. The auxiliary connections are constrained to be of low rank. This not only allows us to rapidly optimize for the new task but also reduces the storage and run-time complexity during the deployment phase. The efficiency of our approach is empirically validated in the stock mid-price movement prediction problem using a large-scale limit order book dataset. Experimental results show that our approach enhances prediction performance as well as reduces the overall number of network parameters.

Mehdi Rafiei, Alexandros Iosifidis

Using a discriminative representation obtained by supervised deep learning methods showed promising results on diverse Content-Based Image Retrieval (CBIR) problems. However, existing methods exploiting labels during training try to discriminate all available classes, which is not ideal in cases where the retrieval problem focuses on a class of interest. In this paper, we propose a regularized loss for Variational Auto-Encoders (VAEs) forcing the model to focus on a given class of interest. As a result, the model learns to discriminate the data belonging to the class of interest from any other possibility, making the learnt latent space of the VAE suitable for class-specific retrieval tasks. The proposed Class-Specific Variational Auto-Encoder (CS-VAE) is evaluated on three public and one custom datasets, and its performance is compared with that of three related VAE-based methods. Experimental results show that the proposed method outperforms its competition in both in-domain and out-of-domain retrieval problems.

Illia Oleksiienko, Dat Thanh Tran, Alexandros Iosifidis

Bayesian Neural Networks (BNNs) provide a tool to estimate the uncertainty of a neural network by considering a distribution over weights and sampling different models for each input. In this paper, we propose a method for uncertainty estimation in neural networks which, instead of considering a distribution over weights, samples outputs of each layer from a corresponding Gaussian distribution, parametrized by the predictions of mean and variance sub-layers. In uncertainty quality estimation experiments, we show that the proposed method achieves better uncertainty quality than other single-bin Bayesian Model Averaging methods, such as Monte Carlo Dropout or Bayes By Backpropagation methods.

Quoc Minh Nguyen, Dat Thanh Tran, Juho Kanniainen et al.

Many cryptocurrency brokers nowadays offer a variety of derivative assets that allow traders to perform hedging or speculation. This paper proposes an effective algorithm based on neural networks to take advantage of these investment products. The proposed algorithm constructs a portfolio that contains a pair of negatively correlated assets. A deep neural network, which outputs the allocation weight of each asset at a time interval, is trained to maximize the Sharpe ratio. A novel loss term is proposed to regulate the network's bias towards a specific asset, thus enforcing the network to learn an allocation strategy that is close to a minimum variance strategy. Extensive experiments were conducted using data collected from Binance spanning 19 months to evaluate the effectiveness of our approach. The backtest results show that the proposed algorithm can produce neural networks that are able to make profits in different market situations.

Morten From Elvebakken, Alexandros Iosifidis, Lukas Esterle

Federated Learning offers a way to train deep neural networks in a distributed fashion. While this addresses limitations related to distributed data, it incurs a communication overhead as the model parameters or gradients need to be exchanged regularly during training. This can be an issue with large scale distribution of learning tasks and negate the benefit of the respective resource distribution. In this paper, we we propose to utilise parallel Adapters for Federated Learning. Using various datasets, we show that Adapters can be incorporated to different Federated Learning techniques. We highlight that our approach can achieve similar inference performance compared to training the full model while reducing the communication overhead by roughly 90%. We further explore the applicability of Adapters in cross-silo and cross-device settings, as well as different non-IID data distributions.

Arian Bakhtiarnia, Błażej Leporowski, Lukas Esterle et al.

Images and video frames captured by cameras placed throughout smart cities are often transmitted over the network to a server to be processed by deep neural networks for various tasks. Transmission of raw images, i.e., without any form of compression, requires high bandwidth and can lead to congestion issues and delays in transmission. The use of lossy image compression techniques can reduce the quality of the images, leading to accuracy degradation. In this paper, we analyze the effect of applying low-overhead lossy image compression methods on the accuracy of visual crowd counting, and measure the trade-off between bandwidth reduction and the obtained accuracy.

Anssi Männistö, Mert Seker, Alexandros Iosifidis et al.

Applying machine learning tools to digitized image archives has a potential to revolutionize quantitative research of visual studies in humanities and social sciences. The ability to process a hundredfold greater number of photos than has been traditionally possible and to analyze them with an extensive set of variables will contribute to deeper insight into the material. Overall, these changes will help to shift the workflow from simple manual tasks to more demanding stages. In this paper, we introduce Automatic Image Content Extraction (AICE) framework for machine learning-based search and analysis of large image archives. We developed the framework in a multidisciplinary research project as framework for future photographic studies by reformulating and expanding the traditional visual content analysis methodologies to be compatible with the current and emerging state-of-the-art machine learning tools and to cover the novel machine learning opportunities for automatic content analysis. The proposed framework can be applied in several domains in humanities and social sciences, and it can be adjusted and scaled into various research settings. We also provide information on the current state of different machine learning techniques and show that there are already various publicly available methods that are suitable to a wide-scale of visual content analysis tasks.

Illia Oleksiienko, Alexandros Iosifidis

Autonomous driving needs to rely on high-quality 3D object detection to ensure safe navigation in the world. Uncertainty estimation is an effective tool to provide statistically accurate predictions, while the associated detection uncertainty can be used to implement a more safe navigation protocol or include the user in the loop. In this paper, we propose a Variational Neural Network-based TANet 3D object detector to generate 3D object detections with uncertainty and introduce these detections to an uncertainty-aware AB3DMOT tracker. This is done by applying a linear transformation to the estimated uncertainty matrix, which is subsequently used as a measurement noise for the adopted Kalman filter. We implement two ways to estimate output uncertainty, i.e., internally, by computing the variance of the CNN outputs and then propagating the uncertainty through the post-processing, and externally, by associating the final predictions of different samples and computing the covariance of each predicted box. In experiments, we show that the external uncertainty estimation leads to better results, outperforming both internal uncertainty estimation and classical tracking approaches. Furthermore, we propose a method to initialize the Variational 3D object detector with a pretrained TANet model, which leads to the best performing models.

Arian Bakhtiarnia, Qi Zhang, Alexandros Iosifidis

Many deep learning tasks require annotations that are too time consuming for human operators, resulting in small dataset sizes. This is especially true for dense regression problems such as crowd counting which requires the location of every person in the image to be annotated. Techniques such as data augmentation and synthetic data generation based on simulations can help in such cases. In this paper, we introduce PromptMix, a method for artificially boosting the size of existing datasets, that can be used to improve the performance of lightweight networks. First, synthetic images are generated in an end-to-end data-driven manner, where text prompts are extracted from existing datasets via an image captioning deep network, and subsequently introduced to text-to-image diffusion models. The generated images are then annotated using one or more high-performing deep networks, and mixed with the real dataset for training the lightweight network. By extensive experiments on five datasets and two tasks, we show that PromptMix can significantly increase the performance of lightweight networks by up to 26%.

Martin Magris, Mostafa Shabani, Alexandros Iosifidis

We develop an optimization algorithm suitable for Bayesian learning in complex models. Our approach relies on natural gradient updates within a general black-box framework for efficient training with limited model-specific derivations. It applies within the class of exponential-family variational posterior distributions, for which we extensively discuss the Gaussian case for which the updates have a rather simple form. Our Quasi Black-box Variational Inference (QBVI) framework is readily applicable to a wide class of Bayesian inference problems and is of simple implementation as the updates of the variational posterior do not involve gradients with respect to the model parameters, nor the prescription of the Fisher information matrix. We develop QBVI under different hypotheses for the posterior covariance matrix, discuss details about its robust and feasible implementation, and provide a number of real-world applications to demonstrate its effectiveness.

Kateryna Chumachenko, Alexandros Iosifidis, Moncef Gabbouj

This paper proposes an approach for improving performance of unimodal models with multimodal training. Our approach involves a multi-branch architecture that incorporates unimodal models with a multimodal transformer-based branch. By co-training these branches, the stronger multimodal branch can transfer its knowledge to the weaker unimodal branches through a multi-task objective, thereby improving the performance of the resulting unimodal models. We evaluate our approach on tasks of dynamic hand gesture recognition based on RGB and Depth, audiovisual emotion recognition based on speech and facial video, and audio-video-text based sentiment analysis. Our approach outperforms the conventionally trained unimodal counterparts. Interestingly, we also observe that optimization of the unimodal branches improves the multimodal branch, compared to a similar multimodal model trained from scratch.

Christian Remi Wewer, Alexandros Iosifidis

Moisture content (MC) estimation is important in the manufacturing process of drying bulky filter media products as it is the prerequisite for drying optimization. In this study, a dataset collected by performing 161 drying industrial experiments is described and a methodology for MC estimation in an non-destructive and online manner during industrial drying is presented. An artificial neural network (ANN) based method is compared to state-of-the-art MC estimation methods reported in the literature. Results of model fitting and training show that a three-layer Perceptron achieves the lowest error. Experimental results show that ANNs combined with oven settings data, drying time and product temperature can be used to reliably estimate the MC of bulky filter media products.

Martin Magris, Mostafa Shabani, Alexandros Iosifidis

We propose an optimization algorithm for Variational Inference (VI) in complex models. Our approach relies on natural gradient updates where the variational space is a Riemann manifold. We develop an efficient algorithm for Gaussian Variational Inference whose updates satisfy the positive definite constraint on the variational covariance matrix. Our Manifold Gaussian Variational Bayes on the Precision matrix (MGVBP) solution provides simple update rules, is straightforward to implement, and the use of the precision matrix parametrization has a significant computational advantage. Due to its black-box nature, MGVBP stands as a ready-to-use solution for VI in complex models. Over five datasets, we empirically validate our feasible approach on different statistical and econometric models, discussing its performance with respect to baseline methods.

Illia Oleksiienko, Alexandros Iosifidis

Deep Ensembles, as a type of Bayesian Neural Networks, can be used to estimate uncertainty on the prediction of multiple neural networks by collecting votes from each network and computing the difference in those predictions. In this paper, we introduce a method for uncertainty estimation that considers a set of independent categorical distributions for each layer of the network, giving many more possible samples with overlapped layers than in the regular Deep Ensembles. We further introduce an optimized inference procedure that reuses common layer outputs, achieving up to 19x speed up and reducing memory usage quadratically. We also show that the method can be further improved by ranking samples, resulting in models that require less memory and time to run while achieving higher uncertainty quality than Deep Ensembles.

Mehdi Rafiei, Alexandros Iosifidis

In anomaly detection, identification of anomalies across diverse product categories is a complex task. This paper introduces a new model by including class discriminative properties obtained by a modified Regularized Discriminative Variational Auto-Encoder (RD-VAE) in the feature extraction process of Coupled-hypersphere-based Feature Adaptation (CFA). By doing so, the proposed Regularized Discriminative Coupled-hypersphere-based Feature Adaptation (RD-CFA), forms a solution for multi-class anomaly detection. By using the discriminative power of RD-VAE to capture intricate class distributions, combined with CFA's robust anomaly detection capability, the proposed method excels in discerning anomalies across various classes. Extensive evaluations on multi-class anomaly detection and localization using the MVTec AD and BeanTech AD datasets showcase the effectiveness of RD-CFA compared to eight leading contemporary methods.

Martin Magris, Alexandros Iosifidis

The Bayesian estimation of GARCH-family models has been typically addressed through Monte Carlo sampling. Variational Inference is gaining popularity and attention as a robust approach for Bayesian inference in complex machine learning models; however, its adoption in econometrics and finance is limited. This paper discusses the extent to which Variational Inference constitutes a reliable and feasible alternative to Monte Carlo sampling for Bayesian inference in GARCH-like models. Through a large-scale experiment involving the constituents of the S&P 500 index, several Variational Inference optimizers, a variety of volatility models, and a case study, we show that Variational Inference is an attractive, remarkably well-calibrated, and competitive method for Bayesian learning.

Arian Bakhtiarnia, Qi Zhang, Alexandros Iosifidis

JPEG image compression algorithm is a widely used technique for image size reduction in edge and cloud computing settings. However, applying such lossy compression on images processed by deep neural networks can lead to significant accuracy degradation. Inspired by the curriculum learning paradigm, we propose a training approach called curriculum pre-training (CPT) for crowd counting on compressed images, which alleviates the drop in accuracy resulting from lossy compression. We verify the effectiveness of our approach by extensive experiments on three crowd counting datasets, two crowd counting DNN models and various levels of compression. The proposed training method is not overly sensitive to hyper-parameters, and reduces the error, particularly for heavily compressed images, by up to 19.70%.

Amir Ramezani Dooraki, Alexandros Iosifidis

Curiosity is one of the main motives in many of the natural creatures with measurable levels of intelligence for exploration and, as a result, more efficient learning. It makes it possible for humans and many animals to explore efficiently by searching for being in states that make them surprised with the goal of learning more about what they do not know. As a result, while being curious, they learn better. In the machine learning literature, curiosity is mostly combined with reinforcement learning-based algorithms as an intrinsic reward. This work proposes an algorithm based on the drive of curiosity for autonomous learning to control by generating proper motor speeds from odometry data. The quadcopter controlled by our proposed algorithm can pass through obstacles while controlling the Yaw direction of the quad-copter toward the desired location. To achieve that, we also propose a new curiosity approach based on prediction error. We ran tests using on-policy, off-policy, on-policy plus curiosity, and the proposed algorithm and visualized the effect of curiosity in evolving exploration patterns. Results show the capability of the proposed algorithm to learn optimal policy and maximize reward where other algorithms fail to do so.

Negar Heidari, Alexandros Iosifidis

Geometric Deep Learning techniques have become a transformative force in the field of Computer-Aided Design (CAD), and have the potential to revolutionize how designers and engineers approach and enhance the design process. By harnessing the power of machine learning-based methods, CAD designers can optimize their workflows, save time and effort while making better informed decisions, and create designs that are both innovative and practical. The ability to process the CAD designs represented by geometric data and to analyze their encoded features enables the identification of similarities among diverse CAD models, the proposition of alternative designs and enhancements, and even the generation of novel design alternatives. This survey offers a comprehensive overview of learning-based methods in computer-aided design across various categories, including similarity analysis and retrieval, 2D and 3D CAD model synthesis, and CAD generation from point clouds, and single/multi-view images. Additionally, it provides a complete list of benchmark datasets and their characteristics, along with open-source codes that have propelled research in this domain. The final discussion delves into the challenges prevalent in this field, followed by potential future research directions in this rapidly evolving field.

Kateryna Chumachenko, Alexandros Iosifidis, Moncef Gabbouj

Dynamic Facial Expression Recognition (DFER) has received significant interest in the recent years dictated by its pivotal role in enabling empathic and human-compatible technologies. Achieving robustness towards in-the-wild data in DFER is particularly important for real-world applications. One of the directions aimed at improving such models is multimodal emotion recognition based on audio and video data. Multimodal learning in DFER increases the model capabilities by leveraging richer, complementary data representations. Within the field of multimodal DFER, recent methods have focused on exploiting advances of self-supervised learning (SSL) for pre-training of strong multimodal encoders. Another line of research has focused on adapting pre-trained static models for DFER. In this work, we propose a different perspective on the problem and investigate the advancement of multimodal DFER performance by adapting SSL-pre-trained disjoint unimodal encoders. We identify main challenges associated with this task, namely, intra-modality adaptation, cross-modal alignment, and temporal adaptation, and propose solutions to each of them. As a result, we demonstrate improvement over current state-of-the-art on two popular DFER benchmarks, namely DFEW and MFAW.

Vidhi Agrawal, Illia Oleksiienko, Alexandros Iosifidis

Dropout is a widely used regularization technique in deep learning, but its effects are typically realized through stochastic masking rather than explicit optimization objectives. We propose a deterministic formulation that expresses dropout as an additive regularizer directly incorporated into the training loss. The framework derives explicit regularization terms for Transformer architectures, covering attention query, key, value, and feed-forward components with independently controllable strengths. This formulation removes reliance on stochastic perturbations while providing clearer and fine-grained control over regularization strength. Experiments across image classification, temporal action detection, and audio classification show that explicit dropout matches or outperforms conventional implicit methods, with consistent gains when applied to attention and feed-forward network layers. Ablation studies demonstrate stable performance and controllable regularization through regularization coefficients and dropout rates. Overall, explicit dropout offers a practical and interpretable alternative to stochastic regularization while maintaining architectural flexibility across diverse tasks.

Johan Andreas Balle Rubak, Khuram Naveed, Sanyam Jain et al.

Objectives: Federated learning (FL) may mitigate privacy constraints, heterogeneous data quality, and inconsistent labeling in dental diagnostic AI. We compared FL with centralized (CL) and local learning (LL) for tooth segmentation in panoramic radiographs across multiple data corruption scenarios. Methods: An Attention U-Net was trained on 2066 radiographs from six institutions across four settings: baseline (unaltered data); label manipulation (dilated/missing annotations); image-quality manipulation (additive Gaussian noise); and exclusion of a faulty client with corrupted data. FL was implemented via the Flower AI framework. Per-client training- and validation-loss trajectories were monitored for anomaly detection and a set of metrics (Dice, IoU, HD, HD95 and ASSD) was evaluated on a hold-out test set. From these metrics significance results were reported through Wilcoxon signed-rank test. CL and LL served as comparators. Results: Baseline: FL achieved a median Dice of 0.94889 (ASSD: 1.33229), slightly better than CL at 0.94706 (ASSD: 1.37074) and LL at 0.93557-0.94026 (ASSD: 1.51910-1.69777). Label manipulation: FL maintained the best median Dice score at 0.94884 (ASSD: 1.46487) versus CL's 0.94183 (ASSD: 1.75738) and LL's 0.93003-0.94026 (ASSD: 1.51910-2.11462). Image noise: FL led with Dice at 0.94853 (ASSD: 1.31088); CL scored 0.94787 (ASSD: 1.36131); LL ranged from 0.93179-0.94026 (ASSD: 1.51910-1.77350). Faulty-client exclusion: FL reached Dice at 0.94790 (ASSD: 1.33113) better than CL's 0.94550 (ASSD: 1.39318). Loss-curve monitoring reliably flagged the corrupted site. Conclusions: FL matches or exceeds CL and outperforms LL across corruption scenarios while preserving privacy. Per-client loss trajectories provide an effective anomaly-detection mechanism and support FL as a practical, privacy-preserving approach for scalable clinical AI deployment.

Muhammad Azlan Qazi, Alexandros Iosifidis, Qi Zhang

Foundation models (FMs) have achieved remarkable success across a wide range of applications, from image classification to natural langurage processing, but pose significant challenges for deployment at edge. This has sparked growing interest in developing practical and efficient strategies for bringing foundation models to edge environments. In this work, we propose PRISM, a communication-efficient and compute-aware strategy for distributed Transformer inference on edge devices. Our method leverages a Segment Means representation to approximate intermediate output features, drastically reducing inter-device communication. Additionally, we restructure the self-attention mechanism to eliminate redundant computations caused by per-device Key/Value calculation in position-wise partitioning and design a partition-aware causal masking scheme tailored for autoregressive models. We evaluate PRISM on ViT, BERT, and GPT-2 across diverse datasets, namely CIFAR-10, CIFAR-100, ImageNet-1k, GLUE, and CBT. Our results demonstrate substantial reductions in communication overhead (up to 99.2% for BERT at compression rate CR = 128) and per-device computation (51.24% for BERT at the same setting), with only minor accuracy degradation. This method offers a scalable and practical solution for deploying foundation models in distributed resource-constrained environments.

Eljas Linna, Kestutis Baltakys, Alexandros Iosifidis et al.

Modeling the dynamics of financial Limit Order Books (LOB) at the message level is challenging due to irregular event timing, rapid regime shifts, and the reactions of high-frequency traders to visible order flow. Previous LOB models require cumbersome data representations and lack adaptability outside their original tasks, leading us to introduce LOBERT, a general-purpose encoder-only foundation model for LOB data suitable for downstream fine-tuning. LOBERT adapts the original BERT architecture for LOB data by using a novel tokenization scheme that treats complete multi-dimensional messages as single tokens while retaining continuous representations of price, volume, and time. With these methods, LOBERT achieves leading performance in tasks such as predicting mid-price movements and next messages, while reducing the required context length compared to previous methods.