Multimodal transformer augmented fusion for speech emotion recognition.

Speech emotion recognition is challenging due to the subjectivity and ambiguity of emotion. In recent years, multimodal methods for speech emotion recognition have achieved promising results. However, due to the heterogeneity of data from different modalities, effectively integrating different modal information remains a difficulty and breakthrough point of the research. Moreover, in view of the limitations of feature-level fusion and decision-level fusion methods, capturing fine-grained modal interactions has often been neglected in previous studies. We propose a method named multimodal transformer augmented fusion that uses a hybrid fusion strategy, combing feature-level fusion and model-level fusion methods, to perform fine-grained information interaction within and between modalities. A Model-fusion module composed of three Cross-Transformer Encoders is proposed to generate multimodal emotional representation for modal guidance and information fusion. Specifically, the multimodal features obtained by feature-level fusion and text features are used to enhance speech features. Our proposed method outperforms existing state-of-the-art approaches on the IEMOCAP and MELD dataset.

Enhanced Neuroactivation during Working Memory Task in Postmenopausal Women Receiving Hormone Therapy: A Coordinate-Based Meta-Analysis.

BACKGROUND AND AIM: Hormone therapy (HT) has long been thought beneficial for controlling menopausal symptoms and human cognition. Studies have suggested that HT has a positive association with working memory, but no consistent relationship between HT and neural activity has been shown in any cognitive domain. The purpose of this meta-analysis was to assess the convergence of findings from published randomized control trials studies that examined brain activation changes in postmenopausal women. METHODS: A systematic search for fMRI studies of neural responses during working memory tasks in postmenopausal women was performed. Studies were excluded if they were not treatment studies and did not contain placebo or blank controls. For the purpose of the meta-analysis, 8 studies were identified, with 103 postmenopausal women taking HT and 109 controls. RESULTS: Compared with controls, postmenopausal women who took HT increased activation in the left frontal lobe, including superior frontal gyrus (BA 8), right middle frontal gyrus (BA 9), anterior lobe, paracentral lobule (BA 7), limbic lobe, and anterior cingulate (BA 32). Additionally, decreased activation is noted in the right limbic lobe, including parahippocampal gyrus (BA 28), left parietal lobe, and superior parietal lobule (BA 7). All regions were significant at p ≤ 0.05 with correction for multiple comparisons. CONCLUSION: Hormone treatment is associated with BOLD signal activation in key anatomical areas during fMRI working memory tasks in healthy hormone-treated postmenopausal women. A positive correlation between activation and task performance suggests that hormone use may benefit working memory.

Fractional anisotropy alterations in individuals born preterm: a diffusion tensor imaging meta-analysis.

AIM: This meta-analysis explored cerebral microstructural changes in individuals born preterm using fractional anisotropy from diffusion tensor imaging. METHOD: We used the activation likelihood estimate (ALE) method for the meta-analysis to locate anatomical regions with white matter abnormalities in a group of individuals born preterm and in term-born comparison participants. A statistical analysis of fractional anisotropy was conducted to quantitatively explore the extent of fractional anisotropy changes in the three subregions of the corpus callosum in the preterm group. RESULTS: ALE analysis identified 11 regions of decreased fractional anisotropy and four regions of increased fractional anisotropy. Analysis of the corpus callosum revealed the largest decrease in fractional anisotropy in the splenium (standardized mean difference [SMD]=-0.75, 95% confidence interval [CI] -0.93 to -0.57), followed by the body (SMD=-0.73, 95% CI -1.13 to -0.32) and the genu (SMD=-0.65, 95% CI -0.97 to -0.33). INTERPRETATION: Significant changes in fractional anisotropy in individuals born preterm reflect white matter abnormalities from childhood to young adulthood, and the mechanism of fractional anisotropy alterations in preterm infants may vary during different stages of white matter development. Furthermore, the variability of fractional anisotropy between studies can primarily be attributed to the age of the individuals at scanning and to the field strength of magnetic resonance scanners.

Generalized Lorenz-Mie theory for a spheroidal particle with off-axis Gaussian-beam illumination.

The beam-shape coefficients of arbitrary off-axis Gaussian beams in spheroidal coordinates are evaluated with a generalized Lorenz-Mie theory. The light-scattering properties of absorbing and nonabsorbing homogeneous spheroidal particles, such as the angular distribution of scattered intensity for a wide range of particles sizes and different complex refractive indices versus the magnitude and location of the beam waist, are investigated.