Browsing by Subject "ICA"
Now showing items 1-2 of 2
-
(2016)Background and aims. Most of the knowledge about neurocognitive processes of reading is based on artificial reading paradigms, such as serial presentation of isolated words or linguistic violation paradigms. The main aim of this thesis was to develop a novel approach to study the neural processes of reading. Specifically, a naturalistic reading task was employed due to concerns for ecological validity, that have been raised about the effects of task on the reading processes. A combination of methods was used to overcome difficulties introduced by this unconstrained reading approach. The second aim was to apply this novel paradigm to test if early differences in the neurocognitive processing of words from different word classes can be found during naturalistic reading. Early processing differences between word classes have been observed before, but they might be task-specific or due to processing related to linguistic violations. Methods. Magnetoencephalography (MEG) and eye movements were recorded simultaneously while participants (8, 4 males) silently read a biographical novel presented on a computer screen. The eye movement recording was used to relate the MEG recording to specific word fixation events during reading. Independent component analysis (ICA) was used to remove eye movement artifacts from the MEG recording and to extract activations of individual cortical areas. An automatic parser was used to extract word class information for all the words in the reading material. Event-related fields (ERFs) evoked by fixations on nouns and verbs were compared using nonparametric cluster-based permutation tests in time window of 0–250 ms after the fixation onset. Results and conclusions. The novel combination of methods used in this study proved to be a promising approach to examine neural processes of reading. In comparison to mainstream methodology of cognitive neuroscience of reading, the present approach has several theoretical and practical advantages. Statistically significant differences between nouns and verbs were found in the sensors above the left temporal cortex, in the 138–164 ms and 184–206 ms time windows after the fixation onset. The results confirm some of the earlier findings that were based on non-naturalistic reading settings and suggests that syntactic and/or semantic information is accessed remarkably early in the course of normal reading.
-
(2021)Independent Component Analysis (ICA) aims to separate the observed signals into their underlying independent components responsible for generating the observations. Most research in ICA has focused on continuous signals, while the methodology for binary and discrete signals is less developed. Yet, binary observations are equally present in various fields and applications, such as causal discovery, signal processing, and bioinformatics. In the last decade, Boolean OR and XOR mixtures have been shown to be identifiable by ICA, but such models suffer from limited expressivity, calling for new methods to solve the problem. In this thesis, "Independent Component Analysis for Binary Data", we estimate the mixing matrix of ICA from binary observations and an additionally observed auxiliary variable by employing a linear model inspired by the Identifiable Variational Autoencoder (iVAE), which exploits the non-stationarity of the data. The model is optimized with a gradient-based algorithm that uses second-order optimization with limited memory, resulting in a training time in the order of seconds for the particular study cases. We investigate which conditions can lead to the reconstruction of the mixing matrix, concluding that the method is able to identify the mixing matrix when the number of observed variables is greater than the number of sources. In such cases, the linear binary iVAE can reconstruct the mixing matrix up to order and scale indeterminacies, which are considered in the evaluation with the Mean Cosine Similarity Score. Furthermore, the model can reconstruct the mixing matrix even under a limited sample size. Therefore, this work demonstrates the potential for applications in real-world data and also offers a possibility to study and formalize identifiability in future work. In summary, the most important contributions of this thesis are the empirical study of the conditions that enable the mixing matrix reconstruction using the binary iVAE, and the empirical results on the performance and efficiency of the model. The latter was achieved through a new combination of existing methods, including modifications and simplifications of a linear binary iVAE model and the optimization of such a model under limited computational resources.
Now showing items 1-2 of 2