Algorithms for Ethical Decision-Making in the Clinic: A Proof of Concept.

Machine intelligence already helps medical staff with a number of tasks. Ethical decision-making, however, has not been handed over to computers. In this proof-of-concept study, we show how an algorithm based on Beauchamp and Childress' prima-facie principles could be employed to advise on a range of moral dilemma situations that occur in medical institutions. We explain why we chose fuzzy cognitive maps to set up the advisory system and how we utilized machine learning to train it. We report on the difficult task of operationalizing the principles of beneficence, non-maleficence and patient autonomy, and describe how we selected suitable input parameters that we extracted from a training dataset of clinical cases. The first performance results are promising, but an algorithmic approach to ethics also comes with several weaknesses and limitations. Should one really entrust the sensitive domain of clinical ethics to machine intelligence?

Efficient Feature Learning Approach for Raw Industrial Vibration Data Using Two-Stage Learning Framework.

In the last decades, data-driven methods have gained great popularity in the industry, supported by state-of-the-art advancements in machine learning. These methods require a large quantity of labeled data, which is difficult to obtain and mostly costly and challenging. To address these challenges, researchers have turned their attention to unsupervised and few-shot learning methods, which produced encouraging results, particularly in the areas of computer vision and natural language processing. With the lack of pretrained models, time series feature learning is still considered as an open area of research. This paper presents an efficient two-stage feature learning approach for anomaly detection in machine processes, based on a prototype few-shot learning technique that requires a limited number of labeled samples. The work is evaluated on a real-world scenario using the publicly available CNC Machining dataset. The proposed method outperforms the conventional prototypical network and the feature analysis shows a high generalization ability achieving an F1-score of 90.3%. The comparison with handcrafted features proves the robustness of the deep features and their invariance to data shifts across machines and time periods, which makes it a reliable method for sensory industrial applications.

Exploring Early Number Abilities With Multimodal Transformers.

Early number skills represent critical milestones in children's cognitive development and are shaped over years of interacting with quantities and numerals in various contexts. Several connectionist computational models have attempted to emulate how certain number concepts may be learned, represented, and processed in the brain. However, these models mainly used highly simplified inputs and focused on limited tasks. We expand on previous work in two directions: First, we train a model end-to-end on video demonstrations in a synthetic environment with multimodal visual and language inputs. Second, we use a more holistic dataset of 35 tasks, covering enumeration, set comparisons, symbolic digits, and seriation. The order in which the model acquires tasks reflects input length and variability, and the resulting trajectories mostly fit with findings from educational psychology. The trained model also displays symbolic and non-symbolic size and distance effects. Using techniques from interpretability research, we investigate how our attention-based model integrates cross-modal representations and binds them into context-specific associative networks to solve different tasks. We compare models trained with and without symbolic inputs and find that the purely non-symbolic model employs more processing-intensive strategies to determine set size.

Comparison of head-related impulse response measurement approaches.

The well-established static head-related transfer function (HRTF) measurement approaches using maximum length sequences and sine sweeps are compared with a recent HRTF estimation approach using normalized least mean square adaptive filters, which enables a continuous movement of the person to be measured during the recording of the excitation signal. By using continuous movement HRTF measurement, a huge amount of time for the individual HRTF estimation can be saved to create a dense HRTF database for headphone-based sound synthesis or applications such as crosstalk cancellation for loudspeaker-based sound synthesis. The different approaches are implemented and experimentally compared by objective and subjective evaluation.

Platelet aggregates detected using quantitative phase imaging associate with COVID-19 severity.

BACKGROUND: The clinical spectrum of acute SARS-CoV-2 infection ranges from an asymptomatic to life-threatening disease. Considering the broad spectrum of severity, reliable biomarkers are required for early risk stratification and prediction of clinical outcomes. Despite numerous efforts, no COVID-19-specific biomarker has been established to guide further diagnostic or even therapeutic approaches, most likely due to insufficient validation, methodical complexity, or economic factors. COVID-19-associated coagulopathy is a hallmark of the disease and is mainly attributed to dysregulated immunothrombosis. This process describes an intricate interplay of platelets, innate immune cells, the coagulation cascade, and the vascular endothelium leading to both micro- and macrothrombotic complications. In this context, increased levels of immunothrombotic components, including platelet and platelet-leukocyte aggregates, have been described and linked to COVID-19 severity. METHODS: Here, we describe a label-free quantitative phase imaging approach, allowing the identification of cell-aggregates and their components at single-cell resolution within 30 min, which prospectively qualifies the method as point-of-care (POC) testing. RESULTS: We find a significant association between the severity of COVID-19 and the amount of platelet and platelet-leukocyte aggregates. Additionally, we observe a linkage between severity, aggregate composition, and size distribution of platelets in aggregates. CONCLUSIONS: This study presents a POC-compatible method for rapid quantitative analysis of blood cell aggregates in patients with COVID-19.

The human body produces a series of immune responses when it gets infected with SARS-CoV-2, the virus that causes COVID-19. One of these responses involves platelets, the blood clotting factor sticking to immune cells to form cell aggregates in the bloodstream. We aimed to understand the significance of these cell aggregates in COVID-19 disease progression. A quantitative imaging approach was used to investigate the number and components of these cell aggregates in SARS-CoV-2 infected patient blood. We observed blood from severe COVID-19 patients was associated with higher numbers and specific composition of cell aggregates. Our method can potentially support the risk stratification of severe patients to prevent complications in COVID-19 and other medical disorders, where immune cells are shown to aggregate.

Low latency localization of multiple sound sources in reverberant environments.

Sound source localization algorithms determine the physical position of a sound source in respect to a listener. For practical applications, a localization algorithm design has to take into account real world conditions like multiple active sources, reverberation, and noise. The application can impose additional constraints on the algorithm, e.g., a requirement for low latency. This work defines the most important constraints for practical applications, introduces an algorithm, which tries to fulfill all requirements as good as possible, and compares it to state-of-the-art sound source localization approaches.

Interpretable PID parameter tuning for control engineering using general dynamic neural networks: An extensive comparison.

Modern automation systems largely rely on closed loop control, wherein a controller interacts with a controlled process via actions, based on observations. These systems are increasingly complex, yet most deployed controllers are linear Proportional-Integral-Derivative (PID) controllers. PID controllers perform well on linear and near-linear systems but their simplicity is at odds with the robustness required to reliably control complex processes. Modern machine learning techniques offer a way to extend PID controllers beyond their linear control capabilities by using neural networks. However, such an extension comes at the cost of losing stability guarantees and controller interpretability. In this paper, we examine the utility of extending PID controllers with recurrent neural networks--namely, General Dynamic Neural Networks (GDNN); we show that GDNN (neural) PID controllers perform well on a range of complex control systems and highlight how they can be a scalable and interpretable option for modern control systems. To do so, we provide an extensive study using four benchmark systems that represent the most common control engineering benchmarks. All control environments are evaluated with and without noise as well as with and without disturbances. The neural PID controller performs better than standard PID control in 15 of 16 tasks and better than model-based control in 13 of 16 tasks. As a second contribution, we address the lack of interpretability that prevents neural networks from being used in real-world control processes. We use bounded-input bounded-output stability analysis to evaluate the parameters suggested by the neural network, making them understandable for engineers. This combination of rigorous evaluation paired with better interpretability is an important step towards the acceptance of neural-network-based control approaches for real-world systems. It is furthermore an important step towards interpretable and safely applied artificial intelligence.

Content adaptive mesh representation of images using binary space partitions.

The interest in content adaptive mesh generation of images has been arising lately due to its wide area of applications in image processing. The major issue is to represent an image with a low number of pixels while preserving its content. These pixels or the nonuniform samples are then used to generate a mesh that approximates the corresponding image. This work presents a novel method based on Binary Space Partitions in combination with three clustering schemes to approximate an image with a mesh. The algorithm has the ability to simultaneously reduce the number of pixels and generate the mesh approximation. The idea is to assume each triangle of the mesh as a plane. Consequently, it will be possible to reconstruct the inlying pixels with planar equations defined from the three nodes of each triangle. If a triangle's equation does not have the ability to reconstruct the pixels lying within up to a predefined error, it is split into two new triangles. Tested on several real images, the proposed method leads to reduced size meshes in a fast manner while retaining the visual quality of the reconstructed images. In addition, it is parallelizable due to the property of Binary Space Partitions which facilitates its application in real-time scenarios.

Step Detection and Parameterization for Gait Assessment Using a Single Waist-Worn Accelerometer.

One of the major reasons why the elderly lose their ability to live independently at home is the decline in gait performance. A measure to assess gait performance using accelerometers is step counting. The main problem with most step detection algorithms is the loss of accuracy at low speeds ( 0.8 m/s) which limits their use in frail elderly populations. In this paper, a step detection algorithm was developed and validated using data from 10 healthy adults and 21 institutionalized seniors, predominantly frail older adults. Data were recorded using a single waist-worn triaxial accelerometer as each of the subjects performed one 10-m-walk trial. The algorithm demonstrated high mean sensitivity (99 ± 1%) for gait speeds between 0.2-1.5 m/s. False positives were evaluated with a series of motion activities performed by one subject. These activities simulate acceleration patterns similar to those generated near the body's center of mass while walking in terms of amplitude signal and periodicity. Cycling was the activity which led to a higher number of false positives. By applying template matching, we reduced by 73% the number of false positives in the cycling activity and eliminated all false positives in the rest of activities. Using K-means clustering, we obtained two different characteristic step patterns, one for normal and one for frail walking, where particular gait events related to limb impacts and muscle flexions were recognized. The proposed system can help to identify seniors at high risk of functional decline and monitor the progress of patients undergoing exercise therapy interventions.

Analysis operator learning and its application to image reconstruction.

Exploiting a priori known structural information lies at the core of many image reconstruction methods that can be stated as inverse problems. The synthesis model, which assumes that images can be decomposed into a linear combination of very few atoms of some dictionary, is now a well established tool for the design of image reconstruction algorithms. An interesting alternative is the analysis model, where the signal is multiplied by an analysis operator and the outcome is assumed to be sparse. This approach has only recently gained increasing interest. The quality of reconstruction methods based on an analysis model severely depends on the right choice of the suitable operator. In this paper, we present an algorithm for learning an analysis operator from training images. Our method is based on l(p)-norm minimization on the set of full rank matrices with normalized columns. We carefully introduce the employed conjugate gradient method on manifolds, and explain the underlying geometry of the constraints. Moreover, we compare our approach to state-of-the-art methods for image denoising, inpainting, and single image super-resolution. Our numerical results show competitive performance of our general approach in all presented applications compared to the specialized state-of-the-art techniques.

Camera-pose estimation via projective Newton optimization on the manifold.

Determining the pose of a moving camera is an important task in computer vision. In this paper, we derive a projective Newton algorithm on the manifold to refine the pose estimate of a camera. The main idea is to benefit from the fact that the 3-D rigid motion is described by the special Euclidean group, which is a Riemannian manifold. The latter is equipped with a tangent space defined by the corresponding Lie algebra. This enables us to compute the optimization direction, i.e., the gradient and the Hessian, at each iteration of the projective Newton scheme on the tangent space of the manifold. Then, the motion is updated by projecting back the variables on the manifold itself. We also derive another version of the algorithm that employs homeomorphic parameterization to the special Euclidean group. We test the algorithm on several simulated and real image data sets. Compared with the standard Newton minimization scheme, we are now able to obtain the full numerical formula of the Hessian with a 60% decrease in computational complexity. Compared with Levenberg-Marquardt, the results obtained are more accurate while having a rather similar complexity.