An important line of research attempts to explain CNN image classifier predictions and intermediate layer representations in terms of human understandable concepts. In this work, we expand on previous works in the literature that use annotated concept datasets to extract interpretable feature space directions and propose an unsupervised post-hoc method to extract a disentangling interpretable basis by looking for the rotation of the feature space that explains sparse one-hot thresholded transformed representations of pixel activations. We do experimentation with existing popular CNNs and demonstrate the effectiveness of our method in extracting an interpretable basis across network architectures and training datasets. We make extensions to the existing basis interpretability metrics found in the literature and show that, intermediate layer representations become more interpretable when transformed to the bases extracted with our method. Finally, using the basis interpretability metrics, we compare the bases extracted with our method with the bases derived with a supervised approach and find that, in one aspect, the proposed unsupervised approach has a strength that constitutes a limitation of the supervised one and give potential directions for future research.

Research exploring CycleGAN-based synthetic image generation has recently accelerated in the medical community due to its ability to leverage unpaired images effectively. However, a commonly established drawback of the CycleGAN, the introduction of artifacts in generated images, makes it unreliable for medical imaging use cases. In an attempt to address this, we explore the effect of structure losses on the CycleGAN and propose a generalized frequency-based loss that aims at preserving the content in the frequency domain. We apply this loss to the use-case of cone-beam computed tomography (CBCT) translation to computed tomography (CT)-like quality. Synthetic CT (sCT) images generated from our methods are compared against baseline CycleGAN along with other existing structure losses proposed in the literature. Our methods (MAE: 85.5, MSE: 20433, NMSE: 0.026, PSNR: 30.02, SSIM: 0.935) quantitatively and qualitatively improve over the baseline CycleGAN (MAE: 88.8, MSE: 24244, NMSE: 0.03, PSNR: 29.37, SSIM: 0.935) across all investigated metrics and are more robust than existing methods. Furthermore, no observable artifacts or loss in image quality were observed. Finally, we demonstrated that sCTs generated using our methods have superior performance compared to the original CBCT images on selected downstream tasks.

In recent years, self-supervised learning (SSL) frameworks have been extensively applied to sensor-based Human Activity Recognition (HAR) in order to learn deep representations without data annotations. While SSL frameworks reach performance almost comparable to supervised models, studies on interpreting representations learnt by SSL models are limited. Nevertheless, modern explainability methods could help to unravel the differences between SSL and supervised representations: how they are being learnt, what properties of input data they preserve, and when SSL can be chosen over supervised training. In this paper, we aim to analyze deep representations of two recent SSL frameworks, namely SimCLR and VICReg. Specifically, the emphasis is made on (i) comparing the robustness of supervised and SSL models to corruptions in input data; (ii) explaining predictions of deep learning models using saliency maps and highlighting what input channels are mostly used for predicting various activities; (iii) exploring properties encoded in SSL and supervised representations using probing. Extensive experiments on two single-device datasets (MobiAct and UCI-HAR) have shown that self-supervised learning representations are significantly more robust to noise in unseen data compared to supervised models. In contrast, features learnt by the supervised approaches are more homogeneous across subjects and better encode the nature of activities.

Emotions play a crucial role in human-human communications with complex socio-psychological nature. In order to enhance emotion communication in human-computer interaction, this paper studies emotion recognition from audio and visual signals in video clips, utilizing facial expressions and vocal utterances. Thereby, the study aims to exploit temporal information of audio-visual cues and detect their informative time segments. Attention mechanisms are used to exploit the importance of each modality over time. We propose a novel framework that consists of bi-modal time windows spanning short video clips labeled with discrete emotions. The framework employs two networks, with each one being dedicated to one modality. As input to a modality-specific network, we consider a time-dependent signal deriving from the embeddings of the video and audio modalities. We employ the encoder part of the Transformer on the visual embeddings and another one on the audio embeddings. The research in this paper introduces detailed studies and meta-analysis findings, linking the outputs of our proposition to research from psychology. Specifically, it presents a framework to understand underlying principles of emotion recognition as functions of three separate setups in terms of modalities: audio only, video only, and the fusion of audio and video. Experimental results on two datasets show that the proposed framework achieves improved accuracy in emotion recognition, compared to state-of-the-art techniques and baseline methods not using attention mechanisms. The proposed method improves the results over baseline methods by at least 5.4%. Our experiments show that attention mechanisms reduce the gap between the entropies of unimodal predictions, which increases the bimodal predictions' certainty and, therefore, improves the bimodal recognition rates. Furthermore, evaluations with noisy data in different scenarios are presented during the training and testing processes to check the framework's consistency and the attention mechanism's behavior. The results demonstrate that attention mechanisms increase the framework's robustness when exposed to similar conditions during the training and the testing phases. Finally, we present comprehensive evaluations of emotion recognition as a function of time. The study shows that the middle time segments of a video clip are essential in the case of using audio modality. However, in the case of video modality, the importance of time windows is distributed equally.

Human Activity Recognition is a field of research where input data can take many forms. Each of the possible input modalities describes human behaviour in a different way, and each has its own strengths and weaknesses. We explore the hypothesis that leveraging multiple modalities can lead to better recognition. Since manual annotation of input data is expensive and time-consuming, the emphasis is made on self-supervised methods which can learn useful feature representations without any ground truth labels. We extend a number of recent contrastive self-supervised approaches for the task of Human Activity Recognition, leveraging inertial and skeleton data. Furthermore, we propose a flexible, general-purpose framework for performing multimodal self-supervised learning, named Contrastive Multiview Coding with Cross-Modal Knowledge Mining (CMC-CMKM). This framework exploits modality-specific knowledge in order to mitigate the limitations of typical self-supervised frameworks. The extensive experiments on two widely-used datasets demonstrate that the suggested framework significantly outperforms contrastive unimodal and multimodal baselines on different scenarios, including fully-supervised fine-tuning, activity retrieval and semi-supervised learning. Furthermore, it shows performance competitive even compared to supervised methods.

Automated Human Activity Recognition has long been a problem of great interest in human-centered and ubiquitous computing. In the last years, a plethora of supervised learning algorithms based on deep neural networks has been suggested to address this problem using various modalities. While every modality has its own limitations, there is one common challenge. Namely, supervised learning requires vast amounts of annotated data which is practically hard to collect. In this paper, we benefit from the self-supervised learning paradigm (SSL) that is typically used to learn deep feature representations from unlabeled data. Moreover, we upgrade a contrastive SSL framework, namely SimCLR, widely used in various applications by introducing a temporal feature alignment procedure for Human Activity Recognition. Specifically, we propose integrating a dynamic time warping (DTW) algorithm in a latent space to force features to be aligned in a temporal dimension. Extensive experiments have been conducted for the unimodal scenario with inertial modality as well as in multimodal settings using inertial and skeleton data. According to the obtained results, the proposed approach has a great potential in learning robust feature representations compared to the recent SSL baselines, and clearly outperforms supervised models in semi-supervised learning. The code for this paper is available via the following link: https://github.com/bulatkh/csshar_tfa.

Applications of AI have found their place in healthcare for improved prediction of disease and associated risks, treatment, and personalization. E-healthcare can substantially improve the quality of life of people with chronic conditions such as neurodegenerative diseases (e.g., Alzheimer's Disease (AD) and Parkinson's Disease (PD)), which are common pathologies in aging people. AD and PD impair senior citizens from performing activities of daily living, increasing the cost of social and health care systems. This paper presents the behavioral analysis module of PROCare4Life, which is an e-healthcare system that focuses on improving the socio-health care of patients with AD and PD. The module generates high-level representations for aspects related to the care of the patients, namely, diseases' symptoms, physical activities, sleep patterns, medication intake, cognitive state, and health records, which are essential variables for further processing in multimodal fusion as part of the decision support system of PROCare4Life. The modular representations are generated from the analysis of long-term patterns of the underlying variables of each category. The proposed modeling benefits from experts' recommendations to improve the monitoring of patients' activities and symptoms.

Wandering is considered to be one of the most common behavioral symptoms of dementia. Designing robust models which are capable of detecting wandering episodes in people living with dementia would allow the prevention of the consequences of this behavior. To tackle this problem, this study proposes a framework where the skeletal trajectories are used to extract patterns from the movements of the participants. These patterns are utilized to classify a movement between wandering and non-wandering behavior using three machine learning methods. The proposed models were assessed based on two datasets collected in different environments consisting of trajectories that are associated with lapping, pacing, and random movements that represent wandering episodes. The predictive model based on the LSTM network achieved the best classification results in terms of macro F1-scores on both datasets with an overall accuracy higher than 70%. The findings of this study present the potential of LSTM-based predictive models in addressing the wandering recognition problem in a real-world scenario with patients suffering from dementia.

In this paper, we propose two novel AR glasses pose estimation algorithms from single infrared images by using 3D point clouds as an intermediate representation. Our first approach "PointsToRotation" is based on a Deep Neural Network alone, whereas our second approach "PointsToPose" is a hybrid model combining Deep Learning and a voting-based mechanism. Our methods utilize a point cloud estimator, which we trained on multi-view infrared images in a semi-supervised manner, generating point clouds based on one image only. We generate a point cloud dataset with our point cloud estimator using the HMDPose dataset, consisting of multi-view infrared images of various AR glasses with the corresponding 6-DoF poses. In comparison to another point cloud-based 6-DoF pose estimation named CloudPose, we achieve an error reduction of around 50%. Compared to a state-of-the-art image-based method, we reduce the pose estimation error by around 96%.

Much of the focus on emotion recognition has gone into the face and voice as expressive channels, whereas bodily expressions of emotions are understudied. Moreover, current studies lack the explainability of computational features of body movements related to emotional expressions. Perceptual research on body parts' movements shows that features related to the arms' movements are correlated the most with human perception of emotions. In this paper, our research aims at presenting an explainable approach for bodily expressed emotion recognition. It utilizes the body joints of the human skeleton, representing them as a graph, which is used in Graph Convolutional Networks (GCNs). We improve the modelling of the GCNs by using spatial attention mechanisms based on body parts, i.e. arms, legs and torso. Our study presents a state-of-the-art explainable approach supported by experimental results on two challenging datasets. Evaluations show that the proposed methodology offers accurate performance and explainable decisions. The methodology demonstrates which body part contributes the most in its inference, showing the significance of arm movements in emotion recognition.

Deep Learning models, applied to a sensor-based Human Activity Recognition task, usually require vast amounts of annotated time-series data to extract robust features. However, annotating signals coming from wearable sensors can be a tedious and, often, not so intuitive process, that requires specialized tools and predefined scenarios, making it an expensive and time-consuming task. This paper combines one of the most recent advances in Self-Supervised Leaning (SSL), namely a SimCLR framework, with a powerful transformer-based encoder to introduce a Contrastive Self-supervised learning approach to Sensor-based Human Activity Recognition (CSSHAR) that learns feature representations from unlabeled sensory data. Extensive experiments conducted on three widely used public datasets have shown that the proposed method outperforms recent SSL models. Moreover, CSSHAR is capable of extracting more robust features than the identical supervised transformer when transferring knowledge from one dataset to another as well as when very limited amounts of annotated data are available.

One of the most significant challenges in Human Activity Recognition using wearable devices is inter-class similarities and subject heterogeneity. These problems lead to the difficulties in constructing robust feature representations that might negatively affect the quality of recognition. This study, for the first time, applies deep triplet networks with various triplet loss functions and mining methods to the Human Activity Recognition task. Moreover, we introduce a novel method for constructing hard triplets by exploiting similarities between subjects performing the same activities using the concept of Hierarchical Triplet Loss. Our deep triplet models are based on the recent state-of-the-art LSTM networks with two attention mechanisms. The extensive experiments conducted in this paper identify important hyperparameters and settings for training deep metric learning models on widely-used open-source Human Activity Recognition datasets. The comparison of the proposed models against the recent benchmark models shows that deep metric learning approach has the potential to improve the quality of recognition. Specifically, at least one of the implemented triplet networks shows the state-of-the-art results for each dataset used in this study, namely PAMAP2, USC-HAD and MHEALTH. Another positive effect of applying deep triplet networks and especially the proposed sampling algorithm is that feature representations are less affected by inter-class similarities and subject heterogeneity issues.

This article proposes a novel study on personality recognition using video data from different scenarios. Our goal is to jointly model nonverbal behavioral cues with contextual information for a robust, multi-scenario, personality recognition system. Therefore, we build a novel multi-stream Convolutional Neural Network (CNN) framework, which considers multiple sources of information. From a given scenario, we extract spatio-temporal motion descriptors from every individual in the scene, spatio-temporal motion descriptors encoding social group dynamics, and proxemics descriptors to encode the interaction with the surrounding context. All the proposed descriptors are mapped to the same feature space facilitating the overall learning effort. Experiments on two public datasets demonstrate the effectiveness of jointly modeling the mutual Person-Context information, outperforming the state-of-the art-results for personality recognition in two different scenarios. Last, we present CNN class activation maps for each personality trait, shedding light on behavioral patterns linked with personality attributes.

Accessing large, manually annotated audio databases in an effort to create robust models for emotion recognition is a notably difficult task, handicapped by the annotation cost and label ambiguities. On the contrary, there are plenty of publicly available datasets for emotion recognition which are based on facial expressivity due to the prevailing role of computer vision in deep learning research, nowadays. Thereby, in the current work, we performed a study on cross-modal transfer knowledge between audio and facial modalities within the emotional context. More concretely, we investigated whether facial information from videos could be used to boost the awareness and the prediction tracking of emotions in audio signals. Our approach was based on a simple hypothesis: that the emotional state’s content of a person’s oral expression correlates with the corresponding facial expressions. Research in the domain of cognitive psychology was affirmative to our hypothesis and suggests that visual information related to emotions fused with the auditory signal is used from humans in a cross-modal integration schema to better understand emotions. In this regard, a method called dacssGAN (which stands for Domain Adaptation Conditional Semi-Supervised Generative Adversarial Networks) is introduced in this work, in an effort to bridge these two inherently different domains. Given as input the source domain (visual data) and some conditional information that is based on inductive conformal prediction, the proposed architecture generates data distributions that are as close as possible to the target domain (audio data). Through experimentation, it is shown that classification performance of an expanded dataset using real audio enhanced with generated samples produced using dacssGAN (50.29% and 48.65%) outperforms the one obtained merely using real audio samples (49.34% and 46.90%) for two publicly available audio–visual emotion datasets.

This paper proposes a novel study on personality recognition using video data from different scenarios. Our goal is to jointly model nonverbal behavioral cues with contextual information for a robust, multi-scenario, personality recognition system. Therefore, we build a novel multi-stream Convolutional Neural Network framework (CNN), which considers multiple sources of information. From a given scenario, we extract spatio-temporal motion descriptors from every individual in the scene, spatio-temporal motion descriptors encoding social group dynamics, and proxemics descriptors to encode the interaction with the surrounding context. All the proposed descriptors are mapped to the same feature space facilitating the overall learning effort. Experiments on two public datasets demonstrate the effectiveness of jointly modeling the mutual Person-Context information, outperforming the state-of-the art-results for personality recognition in two different scenarios. Lastly, we present CNN class activation maps for each personality trait, shedding light on behavioral patterns linked with personality attributes.

The use of recommendation systems has been extensively applied to different fields from the suggestion of multimedia content to other areas such as education. Thereof, the integration of affect-related information becomes a key factor to enhance the user experience and, when it comes to learning, it can be translated into the maximization of knowledge acquisition. The existence of affect-enhanced datasets is imperative when training a recommender learning system. This article introduces a new dataset gathered under realistic conditions, which includes affect-related information of 33 learners from different knowledge background interacting with an e-learning platform. Matrix factorization techniques were evaluated in order to prove the wealthiness of the dataset and to demonstrate that it could potentially replace the adaptation mechanism of the introduced platform and maximizing the learner's knowledge acquisition.

The use of recommendation systems has been extensively applied to different fields from the suggestion of multimedia content to other areas such as education. Thereof, the integration of affect-related information becomes a key factor to enhance the user experience and, when it comes to learning, it can be translated into the maximization of knowledge acquisition. The existence of affect-enhanced datasets is imperative when training a recommender learning system. This article introduces a new dataset gathered under realistic conditions, which includes affect-related information of 33 learners from different knowledge background interacting with an e-learning platform. Matrix factorization techniques were evaluated in order to prove the wealthiness of the dataset and to demonstrate that it could potentially replace the adaptation mechanism of the introduced platform and maximizing the learner's knowledge acquisition.

The use of recommendation systems has been extensively applied to different fields from the suggestion of multimedia content to other areas such as education. Thereof, the integration of affect-related information becomes a key factor to enhance the user experience and, when it comes to learning, it can be translated into the maximization of knowledge acquisition. The existence of affect-enhanced datasets is imperative when training a recommender learning system. This article introduces a new dataset gathered under realistic conditions, which includes affect-related information of 33 learners from different knowledge background interacting with an e-learning platform. Matrix factorization techniques were evaluated in order to prove the wealthiness of the dataset and to demonstrate that it could potentially replace the adaptation mechanism of the introduced platform and maximizing the learner's knowledge acquisition.

Exploiting the multimodal and temporal interaction between audio-visual channels is essential for automatic audio-video emotion recognition (AVER). Modalities’ strength in emotions and time-window of a video-clip could be further utilized through a weighting scheme such as attention mechanism to capture their complementary information. The attention mechanism is a powerful approach for sequence modeling, which can be employed to fuse audio-video cues overtime. We propose a novel framework which consists of biaudio-visual time-windows that span short video-clips labeled with discrete emotions. Attention is used to weigh these time windows for multimodal learning and fusion. Experimental results on two datasets show that the proposed methodology can achieve an enhanced multimodal emotion recognition.

The goal of this work is to investigate the potential of making use of simple activity and motion patterns in a smart environment for approximating personality cues via machine learning techniques. Towards this goal, we present a novel framework for personality recognition, inspired by both Computer Vision and Psychology. Results show a correlation between several behavioral features and personality traits, as well as insights of which type of everyday tasks induce stronger personality display. We experiment with the use of Support Vector Machines, Random Forests and Gaussian Process classification achieving promising predictive ability, related to personality traits. The obtained results show consistency to a good degree, opening the path for applications in psychology, game industry, ambient assisted living, and other fields.