Abstract: During COVID-19, the depression rate has increased dramatically. Young adults are most vulnerable to the mental health effects of the pandemic. Lower-income families have a higher ratio to be diagnosed with depression than the general population, but less access to clinics. This research aims to achieve early depression detection at low cost, large scale, and high accuracy with an interdisciplinary approach by incorporating clinical practices defined by American Psychiatric Association (APA) as well as multimodal AI framework. The proposed approach detected the nine depression symptoms with Natural Language Processing sentiment analysis and a symptom-based Lexicon uniquely designed for young adults. The experiments were conducted on the multimedia survey results from adolescents and young adults and unbiased Twitter communications. The result was further aggregated with the facial emotional cues analyzed by the Convolutional Neural Network on the multimedia survey videos. Five experiments each conducted on 10k data entries reached consistent results with an average accuracy of 88.31%, higher than the existing natural language analysis models. This approach can reach 300+ million daily active Twitter users and is highly accessible by low-income populations to promote early depression detection to raise awareness in adolescents and young adults and reveal complementary cues to assist clinical depression diagnosis.
Abstract: The article presents an ongoing work in research projects such as SCAN4RECO or ARCH, both funded by the European Commission under Horizon 2020 program. The former one concerns multimodal and multispectral scanning of Cultural Heritage assets for their digitization and conservation via spatiotemporal reconstruction and 3D printing, while the latter one aims to better preserve areas of cultural heritage from hazards and risks. It co-creates tools that would help pilot cities to save cultural heritage from the effects of climate change. It develops a disaster risk management framework for assessing and improving the resilience of historic areas to climate change and natural hazards. Tools and methodologies are designed for local authorities and practitioners, urban population, as well as national and international expert communities, aiding authorities in knowledge-aware decision making. In this article we focus on 3D modelling of object geometry using primarily photogrammetric methods to achieve very high model accuracy using consumer types of devices, attractive both to professions and hobbyists alike.
Abstract: Multimodality involves the communication between verbal and visual components in various discourses. A painting represents a form of communication between the artist and the viewer in terms of colors, shades, objects, and the title. This paper aims to present how multimodality can be used to decode the verbal and visual dimensions a painting holds. For that purpose, this study uses Kress and van Leeuwen’s theoretical framework of visual grammar for the analysis of the multimodal semiotic resources of selected paintings of Salvador Dali. This study investigates the visual decoding of the selected paintings of Salvador Dali and analyzing their social and political meanings using Kress and van Leeuwen’s framework of visual grammar. The paper attempts to answer the following questions: 1. How far can multimodality decode the verbal and non-verbal meanings of surrealistic art? 2. How can Kress and van Leeuwen’s theoretical framework of visual grammar be applied to analyze Dali’s paintings? 3. To what extent is Kress and van Leeuwen’s theoretical framework of visual grammar apt to deliver political and social messages of Dali? The paper reached the following findings: the framework’s descriptive tools (representational, interactive, and compositional meanings) can be used to analyze the paintings’ title and their visual elements. Social and political messages were delivered by appropriate usage of color, gesture, vectors, modality, and the way social actors were represented.
Abstract: Sentiment analysis is a very active research topic.
Every day, Facebook, Twitter, Weibo, and other social media,
as well as significant e-commerce websites, generate a massive
amount of comments, which can be used to analyse peoples
opinions or emotions. The existing methods for sentiment analysis
are based mainly on sentiment dictionaries, machine learning, and
deep learning. The first two kinds of methods rely on heavily
sentiment dictionaries or large amounts of labelled data. The third
one overcomes these two problems. So, in this paper, we focus
on the third one. Specifically, we survey various sentiment analysis
methods based on convolutional neural network, recurrent neural
network, long short-term memory, deep neural network, deep belief
network, and memory network. We compare their futures, advantages,
and disadvantages. Also, we point out the main problems of
these methods, which may be worthy of careful studies in the
future. Finally, we also examine the application of deep learning in
multimodal sentiment analysis and aspect-level sentiment analysis.
Abstract: Multimodal image registration is a profoundly complex
task which is why deep learning has been used widely to address it in
recent years. However, two main challenges remain: Firstly, the lack
of ground truth data calls for an unsupervised learning approach,
which leads to the second challenge of defining a feasible loss
function that can compare two images of different modalities to judge
their level of alignment. To avoid this issue altogether we implement a
generative adversarial network consisting of two registration networks
GAB, GBA and two discrimination networks DA, DB connected by
spatial transformation layers. GAB learns to generate a deformation
field which registers an image of the modality B to an image of the
modality A. To do that, it uses the feedback of the discriminator DB
which is learning to judge the quality of alignment of the registered
image B. GBA and DA learn a mapping from modality A to modality
B. Additionally, a cycle-consistency loss is implemented. For this,
both registration networks are employed twice, therefore resulting in
images ˆA, ˆB which were registered to ˜B, ˜A which were registered
to the initial image pair A, B. Thus the resulting and initial images
of the same modality can be easily compared. A dataset of liver
CT and MRI was used to evaluate the quality of our approach and
to compare it against learning and non-learning based registration
algorithms. Our approach leads to dice scores of up to 0.80 ± 0.01
and is therefore comparable to and slightly more successful than
algorithms like SimpleElastix and VoxelMorph.
Abstract: Engagement is one of the most important factors in determining successful outcomes and deep learning in students. Existing approaches to detect student engagement involve periodic human observations that are subject to inter-rater reliability. Our solution uses real-time multimodal multisensor data labeled by objective performance outcomes to infer the engagement of students. The study involves four students with a combined diagnosis of cerebral palsy and a learning disability who took part in a 3-month trial over 59 sessions. Multimodal multisensor data were collected while they participated in a continuous performance test. Eye gaze, electroencephalogram, body pose, and interaction data were used to create a model of student engagement through objective labeling from the continuous performance test outcomes. In order to achieve this, a type of continuous performance test is introduced, the Seek-X type. Nine features were extracted including high-level handpicked compound features. Using leave-one-out cross-validation, a series of different machine learning approaches were evaluated. Overall, the random forest classification approach achieved the best classification results. Using random forest, 93.3% classification for engagement and 42.9% accuracy for disengagement were achieved. We compared these results to outcomes from different models: AdaBoost, decision tree, k-Nearest Neighbor, naïve Bayes, neural network, and support vector machine. We showed that using a multisensor approach achieved higher accuracy than using features from any reduced set of sensors. We found that using high-level handpicked features can improve the classification accuracy in every sensor mode. Our approach is robust to both sensor fallout and occlusions. The single most important sensor feature to the classification of engagement and distraction was shown to be eye gaze. It has been shown that we can accurately predict the level of engagement of students with learning disabilities in a real-time approach that is not subject to inter-rater reliability, human observation or reliant on a single mode of sensor input. This will help teachers design interventions for a heterogeneous group of students, where teachers cannot possibly attend to each of their individual needs. Our approach can be used to identify those with the greatest learning challenges so that all students are supported to reach their full potential.
Abstract: Communication in the modern world is increasingly becoming multimodal due to globalization and the digital space we live in which have remarkably affected how people communicate. Accordingly, Multimodal Discourse Analysis (MDA) is an emerging paradigm in discourse studies with the underlying assumption that other semiotic resources such as images, colours, scientific symbolism, gestures, actions, music and sound, etc. combine with language in order to communicate meaning. One of the effective multimodal media that combines both verbal and non-verbal elements to create meaning is political cartoons. Furthermore, since political and social issues are mirrored in political cartoons, these are regarded as potential objects of discourse analysis since they not only reflect the thoughts of the public but they also have the power to influence them. The aim of this paper is to analyze some selected cartoons on the recognition of Jerusalem as Israel's capital by the American President, Donald Trump, adopting a multimodal approach. More specifically, the present research examines how the various semiotic tools and resources utilized by the cartoonists function in projecting the intended meaning. Ten political cartoons, among a surge of editorial cartoons highlighted by the Anti-Defamation League (ADL) - an international Jewish non-governmental organization based in the United States - as publications in different Arabic-language newspapers in Egypt, Saudi Arabia, UAE, Oman, Iran and UK, were purposively selected for semiotic analysis. These editorial cartoons, all published during 6th–18th December 2017, invariably suggest one theme: Jewish and Israeli domination of the United States. The data were analyzed using the framework of Visual Social Semiotics. In accordance with this methodological framework, the selected visual compositions were analyzed in terms of three aspects of meaning: representational, interactive and compositional. In analyzing the selected cartoons, an interpretative approach is being adopted. This approach prioritizes depth to breadth and enables insightful analyses of the chosen cartoons. The findings of the study reveal that semiotic resources are key elements of political cartoons due to the inherent political communication they convey. It is proved that adequate interpretation of the three aspects of meaning is a prerequisite for understanding the intended meaning of political cartoons. It is recommended that further research should be conducted to provide more insightful analyses of political cartoons from a multimodal perspective.
Abstract: Food is a social and cultural aspect of every individual. Food products, processing, and traditions have been identified as cultural objects carrying history and identity of social groups. Traditional recipes are passed down from one generation to the other, often to strengthen the link with the territory. The paper presents CookIT, a web portal developed to collect Italian traditional recipes related to regional cuisine, with the purpose to disseminate the knowledge of typical Italian recipes and the Mediterranean diet which is a significant part of Italian cuisine. The system designed is completed with multimodal means of browsing and data retrieval. Stored recipes can be retrieved integrating and combining a number of different methods and keys, while the results are displayed using classical styles, such as list and mosaic, and also using maps and graphs, with which users can play using available keys for interaction.
Abstract: With extensive application, the performance of unimodal biometrics systems has to face a diversity of problems such as signal and background noise, distortion, and environment differences. Therefore, multimodal biometric systems are proposed to solve the above stated problems. This paper introduces a bimodal biometric recognition system based on the extracted features of the human palm print and iris. Palm print biometric is fairly a new evolving technology that is used to identify people by their palm features. The iris is a strong competitor together with face and fingerprints for presence in multimodal recognition systems. In this research, we introduced an algorithm to the combination of the palm and iris-extracted features using a texture-based descriptor, the Scale Invariant Feature Transform (SIFT). Since the feature sets are non-homogeneous as features of different biometric modalities are used, these features will be concatenated to form a single feature vector. Particle swarm optimization (PSO) is used as a feature selection technique to reduce the dimensionality of the feature. The proposed algorithm will be applied to the Institute of Technology of Delhi (IITD) database and its performance will be compared with various iris recognition algorithms found in the literature.
Abstract: A major challenge in medical studies, especially those that are longitudinal, is the problem of missing measurements which hinders the effective application of many machine learning algorithms. Furthermore, recent Alzheimer's Disease studies have focused on the delineation of Early Mild Cognitive Impairment (EMCI) and Late Mild Cognitive Impairment (LMCI) from cognitively normal controls (CN) which is essential for developing effective and early treatment methods. To address the aforementioned challenges, this paper explores the potential of using the eXtreme Gradient Boosting (XGBoost) algorithm in handling missing values in multiclass classification. We seek a generalized classification scheme where all prodromal stages of the disease are considered simultaneously in the classification and decision-making processes. Given the large number of subjects (1631) included in this study and in the presence of almost 28% missing values, we investigated the performance of XGBoost on the classification of the four classes of AD, NC, EMCI, and LMCI. Using 10-fold cross validation technique, XGBoost is shown to outperform other state-of-the-art classification algorithms by 3% in terms of accuracy and F-score. Our model achieved an accuracy of 80.52%, a precision of 80.62% and recall of 80.51%, supporting the more natural and promising multiclass classification.
Abstract: Optimization is an important tool in making decisions and in analysing physical systems. In mathematical terms, an optimization problem is the problem of finding the best solution from among the set of all feasible solutions. The paper discusses the Whale Optimization Algorithm (WOA), and its applications in different fields. The algorithm is tested using MATLAB because of its unique and powerful features. The benchmark functions used in WOA algorithm are grouped as: unimodal (F1-F7), multimodal (F8-F13), and fixed-dimension multimodal (F14-F23). Out of these benchmark functions, we show the experimental results for F7, F11, and F19 for different number of iterations. The search space and objective space for the selected function are drawn, and finally, the best solution as well as the best optimal value of the objective function found by WOA is presented. The algorithmic results demonstrate that the WOA performs better than the state-of-the-art meta-heuristic and conventional algorithms.
Abstract: People express emotions through different modalities.
Integration of verbal and non-verbal communication channels creates
a system in which the message is easier to understand. Expanding
the focus to several expression forms can facilitate research on
emotion recognition as well as human-machine interaction. In this
article, the authors present a Polish emotional database composed of
three modalities: facial expressions, body movement and gestures,
and speech. The corpora contains recordings registered in studio
conditions, acted out by 16 professional actors (8 male and 8 female).
The data is labeled with six basic emotions categories, according to
Ekman’s emotion categories. To check the quality of performance,
all recordings are evaluated by experts and volunteers. The database
is available to academic community and might be useful in the study
on audio-visual emotion recognition.
Abstract: In this paper, we presented our innovative way of determining the driving context for a driving assistance system. We invoke the fusion of all parameters that describe the context of the environment, the vehicle and the driver to obtain the driving context. We created a training set that stores driving situation patterns and from which the system consults to determine the driving situation. A machine-learning algorithm predicts the driving situation. The driving situation is an input to the fission process that yields the action that must be implemented when the driver needs to be informed or assisted from the given the driving situation. The action may be directed towards the driver, the vehicle or both. This is an ongoing work whose goal is to offer an alternative driving assistance system for safe driving, green driving and comfortable driving. Here, ontologies are used for knowledge representation.
Abstract: In this paper, we propose a framework to help users to search and retrieve the portions in the lecture video of their interest. This is achieved by temporally segmenting and indexing the lecture video using the topic keywords. We use transcribed text from the video and documents relevant to the video topic extracted from the web for this purpose. The keywords for indexing are found by applying the non-negative matrix factorization (NMF) topic modeling techniques on the web documents. Our proposed technique first creates indices on the transcribed documents using the topic keywords, and these are mapped to the video to find the start and end time of the portions of the video for a particular topic. This time information is stored in the index table along with the topic keyword which is used to retrieve the specific portions of the video for the query provided by the users.
Abstract: In a single case study, we show how a conversation analysis (CA) approach can shed light onto the sequential unfolding of human-robot interaction. Relying on video data, we are able to show that CA allows us to investigate the respective turn-taking systems of humans and a NAO robot in their dialogical dynamics, thus pointing out relevant differences. Our fine grained video analysis points out occurring breakdowns and their overcoming, when humans and a NAO-robot engage in a multimodally uttered multi-party communication during a sports guessing game. Our findings suggest that interdisciplinary work opens up the opportunity to gain new insights into the challenging issues of human robot communication in order to provide resources for developing mechanisms that enable complex human-robot interaction (HRI).
Abstract: The haptic modality has brought a new dimension to human computer interaction by engaging the human sense of touch. However, designing appropriate haptic stimuli, and in particular tactile stimuli, for various applications is still challenging. To tackle this issue, we present an intuitive system that facilitates the authoring of tactile gestures for various applications. The system transforms a hand gesture into a tactile gesture that can be rendering using a home-made haptic jacket. A case study is presented to demonstrate the ability of the system to develop tactile gestures that are recognizable by human subjects. Four tactile gestures are identified and tested to intensify the following four emotional responses: high valence – high arousal, high valence – low arousal, low valence – high arousal, and low valence – low arousal. A usability study with 20 participants demonstrated high correlation between the selected tactile gestures and the intended emotional reaction. Results from this study can be used in a wide spectrum of applications ranging from gaming to interpersonal communication and multimodal simulations.
Abstract: This paper gives an intuition regarding glowworm swarm optimization (GSO) technique to solve dynamic economic dispatch (DED) problems of thermal generating units. The objective of the problem is to schedule optimal power generation of dedicated thermal units over a specific time band. In this study, Glowworm swarm optimization technique enables a swarm of agents to split into subgroup, exhibit simultaneous taxis towards each other and rendezvous at multiple optima (not necessarily equal) of a given multimodal function. The feasibility of the GSO method has been tested on ten-unit-test systems where the power balance constraints, operating limits, valve point effects, and ramp rate limits are taken into account. The results obtained by the proposed technique are compared with other heuristic techniques. The results show that GSO technique is capable of producing better results.
Abstract: Modelling of building processes of a multimodal
freight transportation support information system is discussed based
on modern CASE technologies. Functional efficiencies of ports in
the eastern part of the Black Sea are analyzed taking into account
their ecological, seasonal, resource usage parameters. By resources,
we mean capacities of berths, cranes, automotive transport, as well as
work crews and neighbouring airports. For the purpose of designing
database of computer support system for Managerial (Logistics)
function, using Object-Role Modeling (ORM) tool (NORMA–Natural ORM Architecture) is proposed, after which Entity
Relationship Model (ERM) is generated in automated process.
Software is developed based on Process-Oriented and Service-Oriented architecture, in Visual Studio.NET environment.
Abstract: OPEN_EmoRec_II is an open multimodal corpus with
experimentally induced emotions. In the first half of the experiment,
emotions were induced with standardized picture material and in the
second half during a human-computer interaction (HCI), realized
with a wizard-of-oz design. The induced emotions are based on the
dimensional theory of emotions (valence, arousal and dominance).
These emotional sequences - recorded with multimodal data (facial
reactions, speech, audio and physiological reactions) during a
naturalistic-like HCI-environment one can improve classification
methods on a multimodal level.
This database is the result of an HCI-experiment, for which 30
subjects in total agreed to a publication of their data including the
video material for research purposes*. The now available open
corpus contains sensory signal of: video, audio, physiology (SCL,
respiration, BVP, EMG Corrugator supercilii, EMG Zygomaticus
Major) and facial reactions annotations.
Abstract: OPEN_EmoRec_II is an open multimodal corpus with
experimentally induced emotions. In the first half of the experiment,
emotions were induced with standardized picture material and in the
second half during a human-computer interaction (HCI), realized
with a wizard-of-oz design. The induced emotions are based on the
dimensional theory of emotions (valence, arousal and dominance).
These emotional sequences - recorded with multimodal data (facial
reactions, speech, audio and physiological reactions) during a
naturalistic-like HCI-environment one can improve classification
methods on a multimodal level.
This database is the result of an HCI-experiment, for which 30
subjects in total agreed to a publication of their data including the
video material for research purposes*. The now available open
corpus contains sensory signal of: video, audio, physiology (SCL,
respiration, BVP, EMG Corrugator supercilii, EMG Zygomaticus
Major) and facial reactions annotations.