Unimportant Details Research Articles

Optimizing Steering Angle Prediction in Self-Driving Vehicles Using Evolutionary Convolutional Neural Networks

The global community is awaiting the advent of a self-driving vehicle that is safe, reliable, and capable of navigating a diverse range of road conditions and terrains. This requires a lot of research, study, and optimization. Thus, this work focused on implementing, training, and optimizing a convolutional neural network (CNN) model, aiming to predict the steering angle during driving (one of the main issues). The considered dataset comprises images collected inside a car-driving simulator and further processed for augmentation and removal of unimportant details. In addition, an innovative data-balancing process was previously performed. A CNN model was trained with the dataset, conducting a comparison between several different standard optimizers. Moreover, evolutionary optimization was applied to optimize the model’s weights as well as the optimizers themselves. Several experiments were performed considering different approaches of genetic algorithms (GAs) along with other optimizers from the state of the art. The obtained results demonstrate that the GA is an effective optimization tool for this problem.

Discussion paper: implications for the further development of the successfully in emergency medicine implemented AUD2IT-algorithm.

The AUD2IT-algorithm is a tool to structure the data, which is collected during an emergency treatment. The goal is on the one hand to structure the documentation of the data and on the other hand to give a standardised data structure for the report during handover of an emergency patient. AUD2IT-algorithm was developed to provide residents a documentation aid, which helps to structure the medical reports without getting lost in unimportant details or forgetting important information. The sequence of anamnesis, clinical examination, considering a differential diagnosis, technical diagnostics, interpretation and therapy is rather an academic classification than a description of the real workflow. In a real setting, most of these steps take place simultaneously. Therefore, the application of the AUD2IT-algorithm should also be carried out according to the real processes. A big advantage of the AUD2IT-algorithm is that it can be used as a structure for the entire treatment process and also is entirely usable as a handover protocol within this process to make sure, that the existing state of knowledge is ensured at each point of a team-timeout. PR-E-(AUD2IT)-algorithm makes it possible to document a treatment process that, in principle, does not have to be limited to the field of emergency medicine. Also, in the outpatient treatment the PR-E-(AUD2IT)-algorithm could be used and further developed. One example could be the preparation and allocation of needed resources at the general practitioner. The algorithm is a standardised tool that can be used by healthcare professionals of any level of training. It gives the user a sense of security in their daily work.

‘All These Unimportant Details’: John Ashbery at home

Abstract John Ashbery is a domestic poet. He recognizes that the home contains and organizes a plurality of objects, but also a plurality of thoughts, experiences, and social roles. The domesticity of Ashbery’s poetry materializes many of the contradictory abstractions of modern life, and gives them a place, even if their relationship with each other remains obscure. His poetic homes are microcosms of America, macrocosms of the mind, and material languages. The suburban home is defined in opposition to everything chaotic, decentred, and unstable: rapidly advancing technology, the cyclical renewals of nature, a carnivorous economy, the expanse of air and sky. Paradoxically, the home depends on these opposing forces, even incorporating them into itself, because for Ashbery the domestic is the conscious frame within which things are understood. His image for this kind of domestic perception is the houseboat: homely, comfortable, anchored between nature and civilization, and always in flux.

Contrastive Pre-training with Adversarial Perturbations for Check-In Sequence Representation Learning

A core step of mining human mobility data is to learn accurate representations for user-generated check-in sequences. The learned representations should be able to fully describe the spatial-temporal mobility patterns of users and the high-level semantics of traveling. However, existing check-in sequence representation learning is usually implicitly achieved by end-to-end models designed for specific downstream tasks, resulting in unsatisfactory generalizable abilities and poor performance. Besides, although the sequence representation learning models that follow the contrastive learning pre-training paradigm have achieved breakthroughs in many fields like NLP, they fail to simultaneously consider the unique spatial-temporal characteristics of check-in sequences and need manual adjustments on the data augmentation strategies. So, directly applying them to check-in sequences cannot yield a meaningful pretext task. To this end, in this paper we propose a contrastive pre-training model with adversarial perturbations for check-in sequence representation learning (CACSR). Firstly, we design a novel spatial-temporal augmentation block for disturbing the spatial-temporal features of check-in sequences in the latent space to relieve the stress of designing manual data augmentation strategies. Secondly, to construct an effective contrastive pretext task, we generate “hard” positive and negative pairs for the check-in sequence by adversarial training. These two designs encourage the model to capture the high-level spatial-temporal patterns and semantics of check-in sequences while ignoring the noisy and unimportant details. We demonstrate the effectiveness and versatility of CACSR on two kinds of downstream tasks using three real-world datasets. The results show that our model outperforms both the state-of-the-art pre-training methods and the end-to-end models.

Bio-Inspired Network for Diagnosing Liver Steatosis in Ultrasound Images.

Using ultrasound imaging to diagnose liver steatosis is of great significance for preventing diseases such as cirrhosis and liver cancer. Accurate diagnosis under conditions of low quality, noise and poor resolutions is still a challenging task. Physiological studies have shown that the visual cortex of the biological visual system has selective attention neural mechanisms and feedback regulation of high features to low features. When processing visual information, these cortical regions selectively focus on more sensitive information and ignore unimportant details, which can effectively extract important features from visual information. Inspired by this, we propose a new diagnostic network for hepatic steatosis. In order to simulate the selection mechanism and feedback regulation of the visual cortex in the ventral pathway, it consists of a receptive field feature extraction module, parallel attention module and feedback connection. The receptive field feature extraction module corresponds to the inhibition of the non-classical receptive field of V1 neurons on the classical receptive field. It processes the input image to suppress the unimportant background texture. Two types of attention are adopted in the parallel attention module to process the same visual information and extract different important features for fusion, which improves the overall performance of the model. In addition, we construct a new dataset of fatty liver ultrasound images and validate the proposed model on this dataset. The experimental results show that the network has good performance in terms of sensitivity, specificity and accuracy for the diagnosis of fatty liver disease.

Easy2Hard: Learning to Solve the Intractables From a Synthetic Dataset for Structure-Preserving Image Smoothing.

Image smoothing is a prerequisite for many computer vision and graphics applications. In this article, we raise an intriguing question whether a dataset that semantically describes meaningful structures and unimportant details can facilitate a deep learning model to smooth complex natural images. To answer it, we generate ground-truth labels from easy samples by candidate generation and a screening test and synthesize hard samples in structure-preserving smoothing by blending intricate and multifarious details with the labels. To take full advantage of this dataset, we present a joint edge detection and structure-preserving image smoothing neural network (JESS-Net). Moreover, we propose the distinctive total variation loss as prior knowledge to narrow the gap between synthetic and real data. Experiments on different datasets and real images show clear improvements of our method over the state of the arts in terms of both the image cleanness and structure-preserving ability. Code and dataset are available at https://github.com/YidFeng/Easy2Hard.

Application of Line Integral Convolution Techniques in the Generation of Fine Art Painting Effects

Since the 1980s, nonrealistic drawing techniques have become one of the main focuses of research in computer graphics. In recent years, with the rapid development of computer graphics, nonrealistic drawing techniques have become capable of rendering any digital image into an image with an alternative artistic style. As a result, many scholars have focused on the pencil sketch style. Against this backdrop, research into how to create more realistic art drawings has become a hot topic. Although the rapid development of computer technology has led to advances in art drawing techniques, traditional fine art painting methods are not yet able to simulate realistic pencil sketches in a quite realistic way. In other words, the following issues have been noted. For example, they fail to simulate the gradation of shades and densities of real sketches, the direction of the shadows does not conform to the general rules of sketching, and the results produced are not artistic and realistic. After all, unlike realistic graphics, which seeks a photo-like realism, nonrealistic graphics wants to highlight the message of the image with the help of computer simulations. As a result, nonrealistic graphics often ignores unimportant details in the image. On the other hand, nonrealistic graphics can also simulate an artist’s painting to a certain extent, thus producing an artistically effective image. Many nonrealistic drawing methods have been devised to simulate the styles of oil, watercolour, and ink, but not many algorithms have been developed to generate colour sketches. In order to address this topic, this study improves a fine art drawing generation approach based on the line integral convolution techniques. To be specific, this investigation begins by segmenting the colour images through cluster analysis. Two basic colours are then assigned to each region by calculating the colour variability, and the density of each colour is calculated using a two-tone mapping technique. After that, the line integral convolution technique is applied to generate sketch textures for each of the two basic colour layers, and the two layers are fused to produce the coloured textures. At the same time, a neon transform is used to generate the sketch contour lines. Eventually, the contours are blended with the coloured textures to obtain the coloured sketch effect. The experimental results show that the art painting generation method designed in this paper is able to automatically transform from colour images to coloured pencil drawings, thus improving the efficiency of the process as well as meeting the real-time requirements.

Autoencoder and Partially Impossible Reconstruction Losses.

The generally unsupervised nature of autoencoder models implies that the main training metric is formulated as the error between input images and their corresponding reconstructions. Different reconstruction loss variations and latent space regularizations have been shown to improve model performances depending on the tasks to solve and to induce new desirable properties such as disentanglement. Nevertheless, measuring the success in, or enforcing properties by, the input pixel space is a challenging endeavour. In this work, we want to make use of the available data more efficiently and provide design choices to be considered in the recording or generation of future datasets to implicitly induce desirable properties during training. To this end, we propose a new sampling technique which matches semantically important parts of the image while randomizing the other parts, leading to salient feature extraction and a neglection of unimportant details. The proposed method can be combined with any existing reconstruction loss and the performance gain is superior to the triplet loss. We analyse the resulting properties on various datasets and show improvements on several computer vision tasks: illumination and unwanted features can be normalized or smoothed out and shadows are removed such that classification or other tasks work more reliably; a better invariances with respect to unwanted features is induced; the generalization capacities from synthetic to real images is improved, such that more of the semantics are preserved; uncertainty estimation is superior to Monte Carlo Dropout and an ensemble of models, particularly for datasets of higher visual complexity. Finally, classification accuracy by means of simple linear classifiers in the latent space is improved compared to the triplet loss. For each task, the improvements are highlighted on several datasets commonly used by the research community, as well as in automotive applications.

The summary writing performance of bilingual learners with reading difficulties.

writing is an important tactic for learning from text and the summaries provide information on students' comprehension and learning processes. We investigated the nature of the summaries produced by bilingual adolescents, and whether their summaries were related to their reading abilities in their first and second languages. In each language, we examined the performance of students identified as typically developing, poor decoders, or poor comprehenders. Participants were 246 grade 8 students enrolled in English immersion programs in China. Measures included English word reading and reading comprehension, Chinese word reading and reading comprehension, and nonverbal ability. Students' text-absent summaries of an English 254-word expository passage were analyzed for the number of themes, main ideas, important details, and unimportant details. Using latent profile analysis, participants were identified as typical readers (TR, n = 123), poor decoders (PD, n = 74), or poor comprehenders (PC, n = 49) in English, and TR (n = 129), PD (n = 74), or PC (n = 43) in Chinese, based on word reading and reading comprehension in both English and Chinese. MANCOVA results showed that after controlling for nonverbal ability, in the English-defined groups, the TR group outperformed PD and PC on themes, main ideas, and important details; in the Chinese-defined groups, the TR group outperformed PD and PC on themes, TR performed better than PC on main ideas and important details, and PD outperformed PC on main ideas. Discussion focuses on the difficulties faced by bilingual students with reading difficulties and on the potential of summary writing instruction to improve their comprehension and learning processes.

Pseudo-labeling generative adversarial networks for medical image classification

Semi-supervised learning has become a popular technology in recent years. In this paper, we propose a novel semi-supervised medical image classification algorithm, called Pseudo-Labeling Generative Adversarial Networks (PLGAN), which only uses a small number of real images with few labels to generate fake images or mask images to enlarge the sample size of the labeled training set. First, we combine MixMatch to generate pseudo labels for the fake and unlabeled images to do the classification. Second, contrastive learning and self-attention mechanisms are introduced into PLGAN to exclude the influence of unimportant details. Third, the problem of mode collapse in contrastive learning is well addressed by cyclic consistency loss. Finally, we design global and local classifiers to complement each other with the key information needed for classification. The experimental results on four medical image datasets show that PLGAN can obtain relatively high learning performance by using few labeled and unlabeled data. For example, the classification accuracy of PLGAN is 11% higher than that of MixMatch with 100 labeled images and 1000 unlabeled images on the OCT dataset. In addition, we also conduct other experiments to verify the effectiveness of our algorithm.

Data‐driven polyline simplification using a stacked autoencoder‐based deep neural network

AbstractAutomatic simplification of polylines is an important issue in spatial database and mapping. Traditional rule‐based methods are usually limited in performance, especially when the man‐made rules have to be adapted to different polylines with different shapes and structures. Compared to the existing neural network methods focusing only on the output layer or the code layers for classification or regression, our proposed method generates multi‐level abstractions of polylines by extracting features from multiple hidden layers. Specifically, we first organize the cartographic polylines into the form of feature vectors acceptable to the neural network model. Then, a stacked autoencoder‐based deep neural network model is trained to learn the pattern features of polyline bends and omit unimportant details layer by layer. Finally, the multi‐level abstractions of input polylines are generated from different hidden layers of a single model. The experimental results demonstrate that, compared with the classic Douglas–Peucker and Wang and Muller algorithms, the proposed method is able to properly simplify the polylines while representing their essential shapes smoothly and reducing areal displacement.

CT reconstruction algorithm based on truncated TV

This paper proposes a simultaneous algebraic reconstruction technique (SART) based on truncated total variation (truncated TV), called SART-truncated TV algorithm (SART-TTV). Truncated TV is particularly effective in removing unimportant details while preserving significant edges. Moreover, the truncated TV only penalizes part of the gradient of the image, so it can solve the problem of excessive smoothness caused by the classic TV penalizing larger gradient amplitude. In this paper, the classic Shepp-Logan model is simulated and the results of sparse angle reconstruction are compared and analyzed. The results show that the quality of the reconstructed image obtained by the SART-TTV algorithm is the highest.

Central statistical monitoring of investigator-led clinical trials in oncology.

Investigator-led clinical trials are pragmatic trials that aim to investigate the benefits and harms of treatments in routine clinical practice. These much-needed trials represent the majority of all trials currently conducted. They are however threatened by the rising costs of clinical research, which are in part due to extensive trial monitoring processes that focus on unimportant details. Risk-based quality management focuses, instead, on “things that really matter”. We discuss the role of central statistical monitoring as part of risk-based quality management. We describe the principles of central statistical monitoring, provide examples of its use, and argue that it could help drive down the cost of randomized clinical trials, especially investigator-led trials, whilst improving their quality.

Downsizing Without Downgrading: Approximated Dynamic Time Warping on Nonvolatile Memories

In recent years, time-series data have emerged in a variety of application domains, such as wireless sensor networks and surveillance systems. To identify the similarity between time-series data, the Euclidean distance and its variations are common metrics that quantify the differences between time-series data. However, the Euclidean distance is limited by its inability to elastically shift with the time axis, which motivates the development of dynamic time warping (DTW) algorithms. While DTW algorithms have been proven very useful in diversified applications like speech recognition, their efficacy might be seriously affected by the resolution of the time-series data. However, high-resolution time-series data might take up a gigantic amount of main memory and storage space, which will slow down the DTW analysis procedure. This makes the upscaling of DTW analysis more challenging, especially for in-memory data analytics platforms with limited nonvolatile memory space. In this paper, we propose a strategy to downsample time-series data to significantly reduce their size without seriously affecting the precision of the results obtained by DTW algorithms (downsizing without downgrading). In other words, this paper proposes a technique to remove the unimportant details that are largely ignored by DTW algorithms. The efficacy of the proposed technique is verified by a series of experimental studies, where the results are quite encouraging.

Vox Populi. Listening and Witnessing in German-Jewish War Diaries Written During the Third Reich

Times of crisis, such as war, which are characterized by drastic changes and fractures in daily life, call for reflection and explanation. The diary is a popular genre in this respect and is even described as “ the most appropriate written form in times of crisis”. (Nebel : 1948, p. 6) Not only reflecting, but also witnessing becomes highly important in the context of conflict and war. Diary authors endeavor to collect what seem at first sight to be unimportant details to record what has been said, what (rumors) circulated in their time, etc. These have significant value because they reflect the mood prevailing in the city and express the opinion of the population at that time – the so-called “ vox populi”. Diaries therefore represent a micrological archive of daily life which will be scrutinized by a close reading of German-Jewish war diaries written during the Third Reich. The focus will be on diary entries from Willy Cohn and Walter Tausk from Breslau that start from the beginning of the Second World War (September 1939) and run to the day of their deportation in November 1941. The analysis in the first part will revolve around three main topics : the opinions of ordinary people during war, the availability of information combined with criticism of the newspapers, and the consciousness (or not) of the fate of the Jewish citizens. The second part will reflected on the question of the genre of the diary and whether the diaries discussed here can be seen as war diaries.

A Landauer Formula for Bioelectronic Applications.

Recent electronic transport experiments using metallic contacts attached to proteins identified some “stylized facts”, which contradict conventional wisdom that increasing either the spatial distance between the electrodes or the temperature suppresses conductance exponentially. These include nearly temperature-independent conductance over the protein in the 30 to 300 K range, distance-independent conductance within a single protein in the 1 to 10 nm range and an anomalously large conductance in the 0.1 to 10 nS range. In this paper, we develop a generalization of the low temperature Landauer formula, which can account for the joint effects of tunneling and decoherence and can explain these new experimental findings. We use novel approximations, which greatly simplify the mathematical treatment and allow us to calculate the conductance in terms of a handful macroscopic parameters, instead of the myriads of microscopic parameters describing the details of an atomic level quantum chemical computation. The new approach makes it possible to get predictions for the outcomes of new experiments without relying solely on high performance computing and can distinguish important and unimportant details of the protein structures from the point of view of transport properties.

Pela superfície do mundo: objetos e memória na literatura brasileira contemporânea

Contam os especialistas em falsificação de quadros que não é o rosto ou a mão do retratado que delata o falsário – ali ele terá tomado todas as precauções. Seu erro aparecerá em um detalhe pouco importante, talvez na indumentária ou na curva de uma orelha, onde ele se sentirá mais à vontade e, sem se dar conta, deixará uma assinatura: sua própria pincelada. Abandonando, então, o foco no enredo, ou nos protagonistas das narrativas, procuro pensar a literatura brasileira contemporânea a partir de elementos considerados secundários, quando não irrelevantes: os objetos que compõem a cena. Seus significados, muitas vezes evocadores da memória das personagens, ou dos seus, marcam a situação de homens e mulheres no mundo contemporâneo e revelam algo sobre a perspectiva autoral. Este texto, inspirado em diferentes narrativas e mesmo nas artes visuais, é um breve percurso pelas inquietações que fundamentam esta reflexão.

Prefrontal Theta Oscillations Promote Selective Encoding of Behaviorally Relevant Events.

The ability to capture the most relevant information from everyday experiences without constantly learning unimportant details is vital to survival and mental health. While decreased activity of the medial prefrontal cortex (mPFC) is associated with failed or inflexible encoding of relevant events in the hippocampus, mechanisms used by the mPFC to discern behavioral relevance of events are not clear. To address this question, we chemogenetically activated excitatory neurons in the mPFC of male rats and examined its impact on local network activity and differential associative learning dependent on the hippocampus. Rats were exposed to two neutral stimuli in two environments whose contingency with an aversive stimulus changed systematically across days. Over 2 weeks of differential and reversal learning, theta band activity began to ramp up toward the expected onset of the aversive stimulus, and this ramping activity tracked the subsequent shift of the set (stimulus modality to environment) predictive of the aversive stimulus. With chemogenetic mPFC activation, the ramping activity emerged within a few sessions of differential learning, which paralleled faster learning and stronger correlations between the ramping activity and conditioned responses. Chemogenetic mPFC activity, however, did not affect the adjustment of ramping activity or behavior during reversal learning or set-shifting, suggesting that the faster learning was not because of a general enhancement of attention, sensory, or motor processing. Thus, the dynamics of the mPFC network activation during events provide a relevance-signaling mechanism through which the mPFC exerts executive control over the encoding of those events in the hippocampus.

Did medical curiosity kill the cat?

This article was migrated. The article was marked as recommended. Curiosity is the urge to explore, understand, and discover anything new. Curious doctors act as 'frontier workers' between the humanities and natural sciences. In their efforts towards trying to "understand" both their patients and their complaints, they function as detectives searching for every seemingly unimportant detail in their constant endeavor of solving the mysterious medical puzzle of illness and providing their ailed patients with adequate help. The curiosity to discover my patients' character, to learn their hopes and desires, to explore their past and current social environment is a crucial part of the manifold studied foundation of what is termed medical empathy. Therefore, nurturing curiosity during the course of medical training should be highly desirable. That said, it seems all the more regrettable that medical education often suppresses the development of curiosity rather than nurtures it, often overemphasizing verifiable facts rather than stimulating more complex thought and reflection processes. The importance of medical curiosity, previously neglected as a potent effect and success factor in the medical-therapeutic setting, should no longer be misunderstood or denied its role. We, as physicians and lecturers, have the power to ignite our students' medical curiosity, if we manage to transmit the "spark". This Personal View by a German MD is intended as a passionate plea for more curiosity in medical education.

Region covariance based total variation optimization for structure-texture decomposition

With conventional structure-preserving filters, it is not always easy to remove texture details and preserve important structures from images of high complexity. To enhance the performance of structure-texture decomposition, a new method based on region covariance of simple image features and total variation optimization is proposed in this paper. We first identify texture from structure by patch-based covariance, which shows highly discriminative power for textures. Then, a total variation model built on the joint covariance and gradient is used for structure-preserving smoothing. To overcome the inherent limitation of covariance descriptor in locating main structures, patch shifting based on the variation of the region covariance is introduced. We compare our approach with state-of-the-art structure-preserving decomposition methods and the results show that our approach outperforms them in removing unimportant texture details while preserving main structures. Even for images containing the mixture of high-contrast textures with obscure boarders between them, our approach still can improve the decomposition at few extra cost of computation. Besides better decomposition results and robustness for various types of images, the simplicity of our approach make it easy to implement and adaptable to other applications.