A survey of uncertainty in deep neural networks Gawlikowski, Jakob; Tassi, Cedrique Rovile Njieutcheu; Ali, Mohsin ...
The Artificial intelligence review,
10/2023, Volume:
56, Issue:
Suppl 1
Journal Article
Peer reviewed
Open access
Over the last decade, neural networks have reached almost every field of science and become a crucial part of various real world applications. Due to the increasing spread, confidence in neural ...network predictions has become more and more important. However, basic neural networks do not deliver certainty estimates or suffer from over- or under-confidence, i.e. are badly calibrated. To overcome this, many researchers have been working on understanding and quantifying uncertainty in a neural network’s prediction. As a result, different types and sources of uncertainty have been identified and various approaches to measure and quantify uncertainty in neural networks have been proposed. This work gives a comprehensive overview of uncertainty estimation in neural networks, reviews recent advances in the field, highlights current challenges, and identifies potential research opportunities. It is intended to give anyone interested in uncertainty estimation in neural networks a broad overview and introduction, without presupposing prior knowledge in this field. For that, a comprehensive introduction to the most crucial sources of uncertainty is given and their separation into reducible model uncertainty and irreducible data uncertainty is presented. The modeling of these uncertainties based on deterministic neural networks, Bayesian neural networks (BNNs), ensemble of neural networks, and test-time data augmentation approaches is introduced and different branches of these fields as well as the latest developments are discussed. For a practical application, we discuss different measures of uncertainty, approaches for calibrating neural networks, and give an overview of existing baselines and available implementations. Different examples from the wide spectrum of challenges in the fields of medical image analysis, robotics, and earth observation give an idea of the needs and challenges regarding uncertainties in the practical applications of neural networks. Additionally, the practical limitations of uncertainty quantification methods in neural networks for mission- and safety-critical real world applications are discussed and an outlook on the next steps towards a broader usage of such methods is given.
Full text
Available for:
EMUNI, FIS, FZAB, GEOZS, GIS, IJS, IMTLJ, KILJ, KISLJ, MFDPS, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, SBMB, SBNM, UKNU, UL, UM, UPUK, VKSCE, ZAGLJ
We investigate practical aspects of a recently introduced blind (noncoherent) communication scheme, called modulation on conjugate-reciprocal zeros (MOCZ). MOCZ is suitable for a reliable ...transmission of sporadic and short-packets at ultra-low latency and high spectral efficiency via unknown multipath channels, which are assumed to be static over the receive duration of one packet. The information is modulated on the zeros of the transmitted discrete-time baseband signal's <inline-formula> <tex-math notation="LaTeX">z- </tex-math></inline-formula>transform. Because of ubiquitous impairments between the transmitter and receiver clocks, a carrier frequency offset occurs after down-conversion to the baseband. This results in a common rotation of the zeros. To identify fractional rotations of the base angle in the zero-pattern, we propose an oversampled direct zero-testing decoder to identify the most likely one. Integer rotations correspond to cyclic shifts of the binary message, which we determine by cyclically permutable codes (CPC). Additionally, the embedding of CPCs into cyclic codes, enables additive error-correction which reduces the bit-error-rate tremendously. Furthermore, we exploit the trident structure in the signal's autocorrelation for an energy based detector to estimate timing offsets and the effective channel delay spread. We finally demonstrate how this joint data and channel estimation can be largely improved by receive antenna diversity at low SNR.
Here we describe the isolation of stem cells of the human colonic epithelium. Differential cell surface abundance of ephrin type-B receptor 2 (EPHB2) allows the purification of different cell types ...from human colon mucosa biopsies. The highest EPHB2 surface levels correspond to epithelial colonic cells with the longest telomeres and elevated expression of intestinal stem cell (ISC) marker genes. Moreover, using culturing conditions that recreate the ISC niche, a substantial proportion of EPHB2-high cells can be expanded in vitro as an undifferentiated and multipotent population.
Full text
Available for:
DOBA, IJS, IZUM, KILJ, NUK, PILJ, PNG, SAZU, UILJ, UKNU, UL, UM, UPUK
As it becomes increasingly apparent that 4G will not be able to meet the emerging demands of future mobile communication systems, the question what could make up a 5G system, what are the crucial ...challenges, and what are the key drivers is part of intensive, ongoing discussions. Partly due to the advent of compressive sensing, methods that can optimally exploit sparsity in signals have received tremendous attention in recent years. In this paper, we will describe a variety of scenarios in which signal sparsity arises naturally in 5G wireless systems. Signal sparsity and the associated rich collection of tools and algorithms will thus be a viable source for innovation in 5G wireless system design. We will also describe applications of this sparse signal processing paradigm in Multiple Input Multiple Output random access, cloud radio access networks, compressive channel-source network coding, and embedded security. We will also emphasize an important open problem that may arise in 5G system design, for which sparsity will potentially play a key role in their solution.
AP4 Encodes a c-MYC-Inducible Repressor of p21 Jung, Peter; Menssen, Antje; Mayr, Doris ...
Proceedings of the National Academy of Sciences - PNAS,
09/2008, Volume:
105, Issue:
39
Journal Article
Peer reviewed
Open access
In the majority of human tumors, expression of the c-MYC oncogene becomes constitutive. Here, we report that c-MYC directly regulates the expression of AP4 via CACGTG motifs in the first intron of ...the AP4 gene. Induction of AP4 was required for c-MYC-mediated cell cycle reentry of anti-estrogen arrested breast cancer cells and mitogen-mediated repression of the CDK inhibitor p21. AP4 directly repressed p21 by occupying four CAGCTG motifs in the p21 promoter via its basic region. AP4 levels declined after DNA damage, and ectopic AP4 interfered with p53-mediated cell cycle arrest and sensitized cells to apoptosis induced by DNA damaging agents. AP4 expression blocked induction of p21 by TGF-β in human keratinocytes and interfered with up-regulation of p21 and cell cycle arrest during monoblast differentiation. Notably, AP4 is specifically expressed in colonic progenitor and colorectal carcinoma cells. In conclusion, our results indicate that c-MYC employs AP4 to maintain cells in a proliferative, progenitor-like state.
Full text
Available for:
BFBNIB, NMLJ, NUK, PNG, SAZU, UL, UM, UPUK
The p21 gene encodes a CDK-inhibitor, which is induced by p53 and many other anti-proliferative factors. The mechanism of transcriptional repression of p21 by c-MYC has been a subject of intensive ...study for several years, as it may explain how c-MYC promotes cell cycle progression. Recently, we reported a novel mechanism which allows c-MYC to repress p21: c-MYC triggers a transcriptional cascade by directly inducing the gene encoding the bHLH-LZ transcription factor AP4 (TFAP4), which binds to recognition motifs located in the vicinity of the p21 promoter and mediates transcriptional repression of p21. Thereby, AP4 interferes with induction of p21 via the DNA damage response/p53 or TGF-β/Smad pathways and during differentiation. Intriguingly, the expression patterns of c-MYC and AP4 strictly overlap in colonic epithelium and colorectal cancer. Here we survey the recent findings and discuss the role of AP4 for c-MYC function and its potential application for cancer diagnosis and therapy.
Full text
Available for:
BFBNIB, GIS, IJS, KISLJ, NUK, PNG, UL, UM, UPUK
A photothermal super resolution technique is proposed for an improved inspection of internal defects. To evaluate the potential of the laser-based thermographic technique, an additively manufactured ...stainless steel specimen with closely spaced internal cavities is used. Four different experimental configurations in transmission, reflection, stepwise and continuous scanning are investigated. The applied image post-processing method is based on compressed sensing and makes use of the block sparsity from multiple measurement events. This concerted approach of experimental measurement strategy and numerical optimization enables the resolution of internal defects and outperforms conventional thermographic inspection techniques.
Full text
Available for:
IZUM, KILJ, NUK, PILJ, PNG, SAZU, UL, UM, UPUK
The fifth generation of cellular communication systems is foreseen to enable a multitude of new applications and use cases with very different requirements. A new 5G multi-service air interface needs ...to enhance broadband performance as well as provide new levels of reliability, latency, and supported number of users. In this paper, we focus on the massive Machine Type Communications (mMTC) service within a multi-service air interface. Specifically, we present an overview of different physical and medium access techniques to address the problem of a massive number of access attempts in mMTC and discuss the protocol performance of these solutions in a common evaluation framework.
ABSTRACT
Background and objective
The aim of the Pulmonary Rehabilitation Guidelines (Guidelines) is to provide evidence‐based recommendations for the practice of pulmonary rehabilitation (PR) ...specific to Australian and New Zealand healthcare contexts.
Methods
The Guideline methodology adhered to the Appraisal of Guidelines for Research and Evaluation (AGREE) II criteria. Nine key questions were constructed in accordance with the PICO (Population, Intervention, Comparator, Outcome) format and reviewed by a COPD consumer group for appropriateness. Systematic reviews were undertaken for each question and recommendations made with the strength of each recommendation based on the GRADE (Gradings of Recommendations, Assessment, Development and Evaluation) criteria. The Guidelines were externally reviewed by a panel of experts.
Results
The Guideline panel recommended that patients with mild‐to‐severe COPD should undergo PR to improve quality of life and exercise capacity and to reduce hospital admissions; that PR could be offered in hospital gyms, community centres or at home and could be provided irrespective of the availability of a structured education programme; that PR should be offered to patients with bronchiectasis, interstitial lung disease and pulmonary hypertension, with the latter in specialized centres. The Guideline panel was unable to make recommendations relating to PR programme length beyond 8 weeks, the optimal model for maintenance after PR, or the use of supplemental oxygen during exercise training. The strength of each recommendation and the quality of the evidence are presented in the summary.
Conclusion
The Australian and New Zealand Pulmonary Rehabilitation Guidelines present an evaluation of the evidence for nine PICO questions, with recommendations to provide guidance for clinicians and policymakers.
See article, page 622
Full text
Available for:
FZAB, GIS, IJS, KILJ, NLZOH, NUK, OILJ, SAZU, SBCE, SBMB, UL, UM, UPUK
Block-sparse regularization is already well known in active thermal imaging and is used for multiple-measurement-based inverse problems. The main bottleneck of this method is the choice of ...regularization parameters which differs for each experiment. We show the benefits of using a learned block iterative shrinkage thresholding algorithm (LBISTA) that is able to learn the choice of regularization parameters, without the need to manually select them. In addition, LBISTA enables the determination of a suitable weight matrix to solve the underlying inverse problem. Therefore, in this paper we present LBISTA and compare it with state-of-the-art block iterative shrinkage thresholding using synthetically generated and experimental test data from active thermography for defect reconstruction. Our results show that the use of the learned block-sparse optimization approach provides smaller normalized mean square errors for a small fixed number of iterations. Thus, this allows us to improve the convergence speed and only needs a few iterations to generate accurate defect reconstruction in photothermal super-resolution imaging.
Full text
Available for:
IZUM, KILJ, NUK, PILJ, PNG, SAZU, UL, UM, UPUK
PDF
