Lip reading (LR) is the task of predicting the speech utilizing only the visual information of the speaker. In this work, for the first time, the benefits of alternating between spatiotemporal and spatial convolutions for learning effective features from the LR sequences are studied. In this context, a new learnable module named ALSOS (Alternating Spatiotemporal and Spatial Convolutions) is introduced in the proposed LR system. The ALSOS module consists of spatiotemporal (3D) and spatial (2D) convolutions along with two conversion components (3D-to-2D and 2D-to-3D) providing a sequence-to-sequence-mapping. The designed LR system utilizes the ALSOS module in-between ResNet blocks, as well as Temporal Convolutional Networks (TCNs) in the backend for classification. The whole framework is composed by feedforward convolutional along with residual layers and can be trained end-to-end directly from the image sequences in the word-level LR problem. The ALSOS module can capture spatiotemporal dynamics and can be advantageous in the task of LR when combined with the ResNet topology. Experiments with different combinations of ALSOS with ResNet are performed on a dataset in Greek language simulating a medical support application scenario and on the popular large-scale LRW-500 dataset of English words. Results indicate that the proposed ALSOS module can improve the performance of a LR system. Overall, the insertion of ALSOS module into the ResNet architecture obtained higher classification accuracy since it incorporates the contribution of the temporal information captured at different spatial scales of the framework.

Introduction. Professional identity, professionalism, and professional development are concepts that attract research interest over the years. The concepts are complex, multi-faceted, evolving over time, and varying depending on the pre-primary teachers’ (i.e., preschool educators’) and primary school teachers’ context. The conceptualization by teachers themselves is of particular interest. Goal. Explore the use of artificial intelligence techniques and specifically dimensionality reduction methods, in analyzing in-service teachers’ perceptions of the three concepts as captured in closed-type questions with Likert-scale responses. Methods of the Research. Model Likert-scale responses of closed-type questionnaire as multi-dimensional question space. Apply dimensionality reduction techniques to visualize responses in two dimensions. Translate back question distances into theoretical concepts and identify unexplored topics. Results. The dimensionality reduction disclosed novel correlations between the concepts and their components, not previously reported in published literature, where classical statistical analysis was employed. Conclusions. We confirm the applicability and usefulness of the dimensionality reduction method for the analysis of human responses in closed-type questionnaires. Further, we identify through dimensionality reduction two novel research questions that should be further studied in the future.