Angela Botros

Michele Alberti, Angela Botros, Narayan Schuez et al.

In this work, we investigate the application of trainable and spectrally initializable matrix transformations on the feature maps produced by convolution operations. While previous literature has already demonstrated the possibility of adding static spectral transformations as feature processors, our focus is on more general trainable transforms. We study the transforms in various architectural configurations on four datasets of different nature: from medical (ColorectalHist, HAM10000) and natural (Flowers, ImageNet) images to historical documents (CB55) and handwriting recognition (GPDS). With rigorous experiments that control for the number of parameters and randomness, we show that networks utilizing the introduced matrix transformations outperform vanilla neural networks. The observed accuracy increases by an average of 2.2 across all datasets. In addition, we show that the benefit of spectral initialization leads to significantly faster convergence, as opposed to randomly initialized matrix transformations. The transformations are implemented as auto-differentiable PyTorch modules that can be incorporated into any neural network architecture. The entire code base is open-source.

Angela Botros, Narayan Schütz, Christina Röcke et al.

With growing usage of machine learning algorithms and big data in health applications, digital biomarkers have become an important key feature to ensure the success of those applications. In this paper, we focus on one important use-case, the long-term continuous monitoring of the cognitive ability of older adults. The cognitive ability is a factor both for long-term monitoring of people living alone as well as an outcome in clinical studies. In this work, we propose a new digital biomarker for cognitive abilities based on location eigenbehaviour obtained from contactless ambient sensors. Indoor location information obtained from passive infrared sensors is used to build a location matrix covering several weeks of measurement. Based on the eigenvectors of this matrix, the reconstruction error is calculated for various numbers of used eigenvectors. The reconstruction error is used to predict cognitive ability scores collected at baseline, using linear regression. Additionally, classification of normal versus pathological cognition level is performed using a support-vector-machine. Prediction performance is strong for high levels of cognitive ability, but grows weaker for low levels of cognitive ability. Classification into normal versus pathological cognitive ability level reaches high accuracy with a AUC = 0.94. Due to the unobtrusive method of measurement based on contactless ambient sensors, this digital biomarker of cognitive ability is easily obtainable. The usage of the reconstruction error is a strong digital biomarker for the binary classification and, to a lesser extent, for more detailed prediction of interindividual differences in cognition.

Narayan Schütz, Angela Botros, Michael Single et al.

Sensor technologies are becoming increasingly prevalent in the biomedical field, with applications ranging from telemonitoring of people at risk, to using sensor derived information as objective endpoints in clinical trials. To fully utilize sensor information, signals from distinct sensors often have to be temporally aligned. However, due to imperfect oscillators and significant noise, commonly encountered with biomedical signals, temporal alignment of raw signals is an all but trivial problem, with, to-date, no generally applicable solution. In this work, we present Deep Canonical Correlation Alignment (DCCA), a novel, generally applicable solution for the temporal alignment of raw (biomedical) sensor signals. DCCA allows practitioners to directly align raw signals, from distinct sensors, without requiring deep domain knowledge. On a selection of artificial and real datasets, we demonstrate the performance and utility of DCCA under a variety of conditions. We compare the DCCA algorithm to other warping based methods, DCCA outperforms dynamic time warping and cross correlation based methods by an order of magnitude in terms of alignment error. DCCA performs especially well on almost periodic biomedical signals such as heart-beats and breathing patterns. In comparison to existing approaches, that are not tailored towards raw sensor data, DCCA is not only fast enough to work on signals with billions of data points but also provides automatic filtering and transformation functionalities, allowing it to deal with very noisy and even morphologically distinct signals.