Search

This paper introduces M2M Gen, a multi modal framework for generating background music tailored to Japanese manga. The key challenges in this task are the lack of an available dataset or a baseline. To address these challenges, we propose an automated music generation pipeline that produces background music for an input manga book. Initially, we use the dialogues in a manga to detect scene boundaries and perform emotion classification using the characters faces within a scene. Then, we use GPT4o to translate this low level scene information into a high level music directive. Conditioned on the scene information and the music directive, another instance of GPT 4o generates page level music captions to guide a text to music model. This produces music that is aligned with the mangas evolving narrative. The effectiveness of M2M Gen is confirmed through extensive subjective evaluations, showcasing its capability to generate higher quality, more relevant and consistent music that complements specific scenes when compared to our baselines.

In times of crisis, the prompt and precise classification of disaster-related information shared on social media platforms is crucial for effective disaster response and public safety. During such critical events, individuals use social media to communicate, sharing multimodal textual and visual content. However, due to the significant influx of unfiltered and diverse data, humanitarian organizations face challenges in leveraging this information efficiently. Existing methods for classifying disaster-related content often fail to model users' credibility, emotional context, and social interaction information, which are essential for accurate classification. To address this gap, we propose CrisisSpot, a method that utilizes a Graph-based Neural Network to capture complex relationships between textual and visual modalities, as well as Social Context Features to incorporate user-centric and content-centric information. We also introduce Inverted Dual Embedded Attention (IDEA), which captures both harmonious and contrasting patterns within the data to enhance multimodal interactions and provide richer insights. Additionally, we present TSEqD (Turkey-Syria Earthquake Dataset), a large annotated dataset for a single disaster event, containing 10,352 samples. Through extensive experiments, CrisisSpot demonstrated significant improvements, achieving an average F1-score gain of 9.45% and 5.01% compared to state-of-the-art methods on the publicly available CrisisMMD dataset and the TSEqD dataset, respectively.

Human emotion is a complex phenomenon conveyed and perceived through facial expressions, vocal tones, body language, and physiological signals. Multimodal emotion recognition systems can perform well because they can learn complementary and redundant semantic information from diverse sensors. In real-world scenarios, only a subset of the modalities employed for training may be available at test time. Learning privileged information allows a model to exploit data from additional modalities that are only available during training. SOTA methods for PKD have been proposed to distill information from a teacher model (with privileged modalities) to a student model (without privileged modalities). However, such PKD methods utilize point-to-point matching and do not explicitly capture the relational information. Recently, methods have been proposed to distill the structural information. However, PKD methods based on structural similarity are primarily confined to learning from a single joint teacher representation, which limits their robustness, accuracy, and ability to learn from diverse multimodal sources. In this paper, a multi-teacher PKD (MT-PKDOT) method with self-distillation is introduced to align diverse teacher representations before distilling them to the student. MT-PKDOT employs a structural similarity KD mechanism based on a regularized optimal transport (OT) for distillation. The proposed MT-PKDOT method was validated on the Affwild2 and Biovid datasets. Results indicate that our proposed method can outperform SOTA PKD methods. It improves the visual-only baseline on Biovid data by 5.5%. On the Affwild2 dataset, the proposed method improves 3% and 5% over the visual-only baseline for valence and arousal respectively. Allowing the student to learn from multiple diverse sources is shown to increase the accuracy and implicitly avoids negative transfer to the student model.

Policy optimization methods benefit from a simple and tractable policy parametrization, usually the Gaussian for continuous action spaces. In this paper, we consider a broader policy family that remains tractable: the $q$-exponential family. This family of policies is flexible, allowing the specification of both heavy-tailed policies ($q>1$) and light-tailed policies ($q<1$). This paper examines the interplay between $q$-exponential policies for several actor-critic algorithms conducted on both online and offline problems. We find that heavy-tailed policies are more effective in general and can consistently improve on Gaussian. In particular, we find the Student's t-distribution to be more stable than the Gaussian across settings and that a heavy-tailed $q$-Gaussian for Tsallis Advantage Weighted Actor-Critic consistently performs well in offline benchmark problems. Our code is available at \url{https://github.com/lingweizhu/qexp}., Comment: 26 pages, 13 pages main text, 13 pages appendix

Systems for multimodal emotion recognition (ER) are commonly trained to extract features from different modalities (e.g., visual, audio, and textual) that are combined to predict individual basic emotions. However, compound emotions often occur in real-world scenarios, and the uncertainty of recognizing such complex emotions over diverse modalities is challenging for feature-based models. As an alternative, emerging large language models (LLMs) like BERT and LLaMA can rely on explicit non-verbal cues that may be translated from different non-textual modalities (e.g., audio and visual) into text. Textualization of modalities augments data with emotional cues to help the LLM encode the interconnections between all modalities in a shared text space. In such text-based models, prior knowledge of ER tasks is leveraged to textualize relevant non-verbal cues such as audio tone from vocal expressions, and action unit intensity from facial expressions. Since the pre-trained weights are publicly available for many LLMs, training on large-scale datasets is unnecessary, allowing to fine-tune for downstream tasks such as compound ER (CER). This paper compares the potential of text- and feature-based approaches for compound multimodal ER in videos. Experiments were conducted on the challenging C-EXPR-DB dataset in the wild for CER, and contrasted with results on the MELD dataset for basic ER. Our results indicate that multimodal textualization provides lower accuracy than feature-based models on C-EXPR-DB, where text transcripts are captured in the wild. However, higher accuracy can be achieved when the video data has rich transcripts. Our code is available., Comment: 14 pages, 3 figures, ECCVw 2024

This paper explores the idea of using redirective reconfigurable intelligent surfaces (RedRIS) to overcome many of the challenges associated with the conventional reflective RIS. We develop a framework for jointly optimizing the switching matrix of the lens-type RedRIS ports along with the active precoding matrix at the base station (BS) and the receive scaling factor. A joint non-convex optimization problem is formulated under the minimum mean-square error (MMSE) criterion with the aim to maximize the spectral efficiency of each user. In the single-cell scenario, the optimum active precoding matrix at the multi-antenna BS and the receive scaling factor are found in closed-form by applying Lagrange optimization, while the optimal switching matrix of the lens-type RedRIS is obtained by means of a newly developed alternating optimization algorithm. We then extend the framework to the multi-cell scenario with single-antenna base stations that are aided by the same lens-type RedRIS. We further present two methods for reducing the number of effective connections of the RedRIS ports that result in appreciable overhead savings while enhancing the robustness of the system. The proposed RedRIS-based schemes are gauged against conventional reflective RIS-aided systems under both perfect and imperfect channel state information (CSI). The simulation results show the superiority of the proposed schemes in terms of overall throughput while incurring much less control overhead.

Multimodal emotion recognition (MMER) systems typically outperform unimodal systems by leveraging the inter- and intra-modal relationships between, e.g., visual, textual, physiological, and auditory modalities. This paper proposes an MMER method that relies on a joint multimodal transformer (JMT) for fusion with key-based cross-attention. This framework can exploit the complementary nature of diverse modalities to improve predictive accuracy. Separate backbones capture intra-modal spatiotemporal dependencies within each modality over video sequences. Subsequently, our JMT fusion architecture integrates the individual modality embeddings, allowing the model to effectively capture inter- and intra-modal relationships. Extensive experiments on two challenging expression recognition tasks -- (1) dimensional emotion recognition on the Affwild2 dataset (with face and voice) and (2) pain estimation on the Biovid dataset (with face and biosensors) -- indicate that our JMT fusion can provide a cost-effective solution for MMER. Empirical results show that MMER systems with our proposed fusion allow us to outperform relevant baseline and state-of-the-art methods., Comment: 10 pages, 4 figures, 6 tables, CVPRw 2024

Accurate segmentation of multiple organs in Computed Tomography (CT) images plays a vital role in computer-aided diagnosis systems. Various supervised-learning approaches have been proposed recently. However, these methods heavily depend on a large amount of high-quality labeled data, which is expensive to obtain in practice. In this study, we present a label-efficient learning approach using a pre-trained diffusion model for multi-organ segmentation tasks in CT images. First, a denoising diffusion model was trained using unlabeled CT data, generating additional two-dimensional (2D) CT images. Then the pre-trained denoising diffusion network was transferred to the downstream multi-organ segmentation task, effectively creating a semi-supervised learning model that requires only a small amount of labeled data. Furthermore, linear classification and fine-tuning decoder strategies were employed to enhance the network's segmentation performance. Our generative model at 256x256 resolution achieves impressive performance in terms of Fr\'echet inception distance, spatial Fr\'echet inception distance, and F1-score, with values of 11.32, 46.93, and 73.1\%, respectively. These results affirm the diffusion model's ability to generate diverse and realistic 2D CT images. Additionally, our method achieves competitive multi-organ segmentation performance compared to state-of-the-art methods on the FLARE 2022 dataset, particularly in limited labeled data scenarios. Remarkably, even with only 1\% and 10\% labeled data, our method achieves Dice similarity coefficients (DSCs) of 71.56\% and 78.51\% after fine-tuning, respectively. The method achieves a DSC score of 51.81\% using just four labeled CT scans. These results demonstrate the efficacy of our approach in overcoming the limitations of supervised learning heavily reliant on large-scale labeled data.

Financial exchanges are migrating to the cloud, but the best-effort nature of the public cloud is at odds with the stringent latency requirements of exchanges. We present Jasper, a system for meeting the networking requirements of financial exchanges on the public cloud. Jasper uses an overlay tree to scalably multicast market data from an exchange to ~1000 participants with low latency (250 microseconds) and a 1-microsecond difference in data reception time between any two participants. Jasper reuses the same tree for scalable inbound communication (participants to exchange), augmenting it with order pacing and a new priority queue, Limit Order Queue (LOQ), to efficiently handle bursts of market orders. Jasper achieves better scalability and 50% lower latency than the AWS multicast service. During bursty market activity, LOQ nearly doubles the order processing rate.

Emotion recognition (ER) technology is an integral part for developing innovative applications such as drowsiness detection and health monitoring that plays a pivotal role in contemporary society. This study delves into ER using electroencephalography (EEG), within immersive virtual reality (VR) environments. There are four main stages in our proposed methodology including data acquisition, pre-processing, feature extraction, and emotion classification. Acknowledging the limitations of existing 2D datasets, we introduce a groundbreaking 3D VR dataset to elevate the precision of emotion elicitation. Leveraging the Interaxon Muse headband for EEG recording and Oculus Quest 2 for VR stimuli, we meticulously recorded data from 40 participants, prioritizing subjects without reported mental illnesses. Pre-processing entails rigorous cleaning, uniform truncation, and the application of a Savitzky-Golay filter to the EEG data. Feature extraction encompasses a comprehensive analysis of metrics such as power spectral density, correlation, rational and divisional asymmetry, and power spectrum. To ensure the robustness of our model, we employed a 10-fold cross-validation, revealing an average validation accuracy of 85.54\%, with a noteworthy maximum accuracy of 90.20\% in the best fold. Subsequently, the trained model demonstrated a commendable test accuracy of 82.03\%, promising favorable outcomes.

Deep learning models for multimodal expression recognition have reached remarkable performance in controlled laboratory environments because of their ability to learn complementary and redundant semantic information. However, these models struggle in the wild, mainly because of the unavailability and quality of modalities used for training. In practice, only a subset of the training-time modalities may be available at test time. Learning with privileged information enables models to exploit data from additional modalities that are only available during training. State-of-the-art knowledge distillation (KD) methods have been proposed to distill information from multiple teacher models (each trained on a modality) to a common student model. These privileged KD methods typically utilize point-to-point matching, yet have no explicit mechanism to capture the structural information in the teacher representation space formed by introducing the privileged modality. Experiments were performed on two challenging problems - pain estimation on the Biovid dataset (ordinal classification) and arousal-valance prediction on the Affwild2 dataset (regression). Results show that our proposed method can outperform state-of-the-art privileged KD methods on these problems. The diversity among modalities and fusion architectures indicates that PKDOT is modality- and model-agnostic.

The reliable diagnosis of cardiac conditions through electrocardiogram (ECG) analysis critically depends on accurately detecting P waves and measuring the PR interval. However, achieving consistent and generalizable diagnoses across diverse populations presents challenges due to the inherent global variations observed in ECG signals. This paper is focused on applying the Q learning reinforcement algorithm to the various ECG datasets available in the PhysioNet/Computing in Cardiology Challenge (CinC). Five ECG beats, including Normal Sinus Rhythm, Atrial Flutter, Atrial Fibrillation, 1st Degree Atrioventricular Block, and Left Atrial Enlargement, are included to study variations of P waves and PR Interval on Lead II and Lead V1. Q-Agent classified 71,672 beat samples in 8,867 patients with an average accuracy of 90.4% and only 9.6% average hamming loss over misclassification. The average classification time at the 100th episode containing around 40,000 samples is 0.04 seconds. An average training reward of 344.05 is achieved at an alpha, gamma, and SoftMax temperature rate of 0.001, 0.9, and 0.1, respectively.

Adapting a deep learning model to a specific target individual is a challenging facial expression recognition (FER) task that may be achieved using unsupervised domain adaptation (UDA) methods. Although several UDA methods have been proposed to adapt deep FER models across source and target data sets, multiple subject-specific source domains are needed to accurately represent the intra- and inter-person variability in subject-based adaption. This paper considers the setting where domains correspond to individuals, not entire datasets. Unlike UDA, multi-source domain adaptation (MSDA) methods can leverage multiple source datasets to improve the accuracy and robustness of the target model. However, previous methods for MSDA adapt image classification models across datasets and do not scale well to a more significant number of source domains. This paper introduces a new MSDA method for subject-based domain adaptation in FER. It efficiently leverages information from multiple source subjects (labeled source domain data) to adapt a deep FER model to a single target individual (unlabeled target domain data). During adaptation, our subject-based MSDA first computes a between-source discrepancy loss to mitigate the domain shift among data from several source subjects. Then, a new strategy is employed to generate augmented confident pseudo-labels for the target subject, allowing a reduction in the domain shift between source and target subjects. Experiments performed on the challenging BioVid heat and pain dataset with 87 subjects and the UNBC-McMaster shoulder pain dataset with 25 subjects show that our subject-based MSDA can outperform state-of-the-art methods yet scale well to multiple subject-based source domains.

Dynamic susceptibility contrast magnetic resonance imaging (DSC-MRI) is widely used to evaluate acute ischemic stroke to distinguish salvageable tissue and infarct core. For this purpose, traditional methods employ deconvolution techniques, like singular value decomposition, which are known to be vulnerable to noise, potentially distorting the derived perfusion parameters. However, deep learning technology could leverage it, which can accurately estimate clinical perfusion parameters compared to traditional clinical approaches. Therefore, this study presents a perfusion parameters estimation network that considers spatial and temporal information, the Spatiotemporal Network (ST-Net), for the first time. The proposed network comprises a designed physical loss function to enhance model performance further. The results indicate that the network can accurately estimate perfusion parameters, including cerebral blood volume (CBV), cerebral blood flow (CBF), and time to maximum of the residual function (Tmax). The structural similarity index (SSIM) mean values for CBV, CBF, and Tmax parameters were 0.952, 0.943, and 0.863, respectively. The DICE score for the hypo-perfused region reached 0.859, demonstrating high consistency. The proposed model also maintains time efficiency, closely approaching the performance of commercial gold-standard software.

In this paper we investigate the use of reinforcement-learning based prediction approaches for a real drinking-water treatment plant. Developing such a prediction system is a critical step on the path to optimizing and automating water treatment. Before that, there are many questions to answer about the predictability of the data, suitable neural network architectures, how to overcome partial observability and more. We first describe this dataset, and highlight challenges with seasonality, nonstationarity, partial observability, and heterogeneity across sensors and operation modes of the plant. We then describe General Value Function (GVF) predictions -- discounted cumulative sums of observations -- and highlight why they might be preferable to classical n-step predictions common in time series prediction. We discuss how to use offline data to appropriately pre-train our temporal difference learning (TD) agents that learn these GVF predictions, including how to select hyperparameters for online fine-tuning in deployment. We find that the TD-prediction agent obtains an overall lower normalized mean-squared error than the n-step prediction agent. Finally, we show the importance of learning in deployment, by comparing a TD agent trained purely offline with no online updating to a TD agent that learns online. This final result is one of the first to motivate the importance of adapting predictions in real-time, for non-stationary high-volume systems in the real world., Comment: Published in Machine Learning (2023)