521 research outputs found
DeepFloat: Resource-Efficient Dynamic Management of Vehicular Floating Content
No full textOpportunistic communications are expected to playa crucial role in enabling context-aware vehicular services. Awidely investigated opportunistic communication paradigm forstoring a piece of content probabilistically in a geographicalarea is Floating Content (FC). A key issue in the practicaldeployment of FC is how to tune content replication and cachingin a way which achieves a target performance (in terms ofthe mean fraction of users possessing the content in a givenregion of space) while minimizing the use of bandwidth andhost memory. Fully distributed, distance-based approaches provehighly inefficient, and may not meet the performance target,while centralized, model-based approaches do not perform wellin realistic, inhomogeneous settings.In this work, we present a data-driven centralized approachto resource-efficient, QoS-aware dynamic management of FC.We propose a Deep Learning strategy, which employs a Con-volutional Neural Network (CNN) to capture the relationshipsbetween patterns of users mobility, of content diffusion andreplication, and FC performance in terms of resource utilizationand of content availability within a given area. Numericalevaluations show the effectiveness of our approach in derivingstrategies which efficiently modulate the FC operation in spaceand effectively adapt to mobility pattern changes over time
A Deep Learning Strategy for Vehicular Floating Content Management
No full textFloating Content (FC) is a communication paradigm for the local dissemination of contextualized information through D2D connectivity, in a way which minimizes the use of resources while achieving some specified performance target. Existing approaches to FC dimensioning are based on unrealistic system assumptions that make them, highly inaccurate and overly conservative when applied in realistic settings. In this paper, we present a first step towards the development of a cognitive approach to efficient dynamic management of FC. We propose a deep learning strategy for FC dimensioning, which exploits a Convolutional Neural Network (CNN) to efficiently modulate over time the resources employed by FC in a QoS-aware manner. Numerical evaluations show that our approach achieves a maximum rejection rate of 3%, and resource savings of 37.5% with respect to the benchmark strategy
Combining weakly and strongly supervised learning improves strong supervision in Gleason pattern classification
No full textBackground: One challenge to train deep convolutional neural network (CNNs) models with whole slide images (WSIs) is providing the required large number of costly, manually annotated image regions. Strategies to alleviate the scarcity of annotated data include: using transfer learning, data augmentation and training the models with less expensive image-level annotations (weakly-supervised learning). However, it is not clear how to combine the use of transfer learning in a CNN model when different data sources are available for training or how to leverage from the combination of large amounts of weakly annotated images with a set of local region annotations. This paper aims to evaluate CNN training strategies based on transfer learning to leverage the combination of weak and strong annotations in heterogeneous data sources. The trade-off between classification performance and annotation effort is explored by evaluating a CNN that learns from strong labels (region annotations) and is later fine-tuned on a dataset with less expensive weak (image-level) labels. Results: As expected, the model performance on strongly annotated data steadily increases as the percentage of strong annotations that are used increases, reaching a performance comparable to pathologists (κ= 0.691 ± 0.02). Nevertheless, the performance sharply decreases when applied for the WSI classification scenario with κ= 0.307 ± 0.133. Moreover, it only provides a lower performance regardless of the number of annotations used. The model performance increases when fine-tuning the model for the task of Gleason scoring with the weak WSI labels κ= 0.528 ± 0.05. Conclusion: Combining weak and strong supervision improves strong supervision in classification of Gleason patterns using tissue microarrays (TMA) and WSI regions. Our results contribute very good strategies for training CNN models combining few annotated data and heterogeneous data sources. The performance increases in the controlled TMA scenario with the number of annotations used to train the model. Nevertheless, the performance is hindered when the trained TMA model is applied directly to the more challenging WSI classification problem. This demonstrates that a good pre-trained model for prostate cancer TMA image classification may lead to the best downstream model if fine-tuned on the WSI target dataset. We have made available the source code repository for reproducing the experiments in the paper: https://github.com/ilmaro8/Digital_Pathology_Transfer_Learnin
Classification of Noisy Free-Text Prostate Cancer Pathology Reports Using Natural Language Processing
No full textFree-text reporting has been the main approach in clinical pathology practice for decades. Pathology reports are an essential information source to guide the treatment of cancer patients and for cancer registries, which process high volumes of free-text reports annually. Information coding and extraction are usually performed manually and it is an expensive and time-consuming process, since reports vary widely between institutions, usually contain noise and do not have a standard structure. This paper presents strategies based on natural language processing (NLP) models to classify noisy free-text pathology reports of high and low-grade prostate cancer from the open-source repository TCGA (The Cancer Genome Atlas). We used paragraph vectors to encode the reports and compared them with n-grams and TF-IDF representations. The best representation based on distributed bag of words of paragraph vectors obtained an f1 -score of 0.858 and an AUC of 0.854 using a logistic regression classifier. We investigate the classifier’s more relevant words in each case using the LIME interpretability tool, confirming the classifiers’ usefulness to select relevant diagnostic words. Our results show the feasibility of using paragraph embeddings to represent and classify pathology reports
Staining invariant features for improving generalization of deep convolutional neural networks in computational pathology
No full textOne of the main obstacles for the implementation of deep convolutional neural networks (DCNNs) in the clinical pathology workflow is their low capability to overcome variability in slide preparation and scanner configuration, that leads to changes in tissue appearance. Some of these variations may not be not included in the training data, which means that the models have a risk to not generalize well. Addressing such variations and evaluating them in reproducible scenarios allows understanding of when the models generalize better, which is crucial for performance improvements and better DCNN models. Staining normalization techniques (often based on color deconvolution and deep learning) and color augmentation approaches have shown improvements in the generalization of the classification tasks for several tissue types. Domain-invariant training of DCNN's is also a promising technique to address the problem of training a single model for different domains, since it includes the source domain information to guide the training toward domain-invariant features, achieving state-of-the-art results in classification tasks. In this article, deep domain adaptation in convolutional networks (DANN) is applied to computational pathology and compared with widely used staining normalization and color augmentation methods in two challenging classification tasks. The classification tasks rely on two openly accessible datasets, targeting Gleason grading in prostate cancer, and mitosis classification in breast tissue. The benchmark of the different techniques and their combination in two DCNN architectures allows us to assess the generalization abilities and advantages of each method in the considered classification tasks. The code for reproducing our experiments and preprocessing the data is publicly available1. Quantitative and qualitative results show that the use of DANN helps model generalization to external datasets. The combination of several techniques to manage color heterogeneity suggests that several methods together, such as color augmentation methods with DANN training, can generalize even further. The results do not show a single best technique among the considered methods, even when combining them. However, color augmentation and DANN training obtain most often the best results (alone or combined with color normalization and color augmentation). The statistical significance of the results and the embeddings visualizations provide useful insights to design DCNN that generalizes to unseen staining appearances. Furthermore, in this work, we release for the first time code for DANN evaluation in open access datasets for computational pathology. This work opens the possibility for further research on using DANN models together with techniques that can overcome the tissue preparation differences across datasets to tackle limited generalization
Replication Data for: Data policies of highly-ranked social science journals
No full textBy encouraging and requiring that authors share their data in order to publish articles, scholarly journals have become an important actor in the movement to improve the openness of data and the reproducibility of research. But how many social science journals encourage or mandate that authors share the data supporting their research findings? How does the share of journal data policies vary by discipline? What influences these journals’ decisions to adopt such policies and instructions? And what do those policies and instructions look like?
We discuss the results of our analysis of the instructions and policies of 291 highly-ranked journals publishing social science research, where we studied the contents of journal data policies and instructions across 14 variables, such as when and how authors are asked to share their data, and what role journal ranking and age play in the existence and quality of data policies and instructions. We also attempt to compare our results to the results of other studies that have analyzed the policies of social science journals, although differences in the journals chosen and how each study defines what constitutes a data policy limit this comparison.
We conclude that a little more than half of the journals in our study have data policies. A greater share of the economics journals have data policies and mandate sharing, followed by political science/international relations and psychology journals.
Finally, we use our findings to make several recommendations: Policies should include the terms “data”, “dataset” or more specific terms that make it clear what to make available; policies should include the benefits of data sharing; journals, publishers, and associations need to collaborate more to clarify data policies; and policies should explicitly ask for qualitative data. </p
Replication Data for: Data policies of highly-ranked social science journals
No full textBy encouraging and requiring that authors share their data in order to publish articles, scholarly journals have become an important actor in the movement to improve the openness of data and the reproducibility of research. But how many social science journals encourage or mandate that authors share the data supporting their research findings? How does the share of journal data policies vary by discipline? What influences these journals’ decisions to adopt such policies and instructions? And what do those policies and instructions look like?
We discuss the results of our analysis of the instructions and policies of 291 highly-ranked journals publishing social science research, where we studied the contents of journal data policies and instructions across 14 variables, such as when and how authors are asked to share their data, and what role journal ranking and age play in the existence and quality of data policies and instructions. We also attempt to compare our results to the results of other studies that have analyzed the policies of social science journals, although differences in the journals chosen and how each study defines what constitutes a data policy limit this comparison.
We conclude that a little more than half of the journals in our study have data policies. A greater share of the economics journals have data policies and mandate sharing, followed by political science/international relations and psychology journals.
Finally, we use our findings to make several recommendations: Policies should include the terms “data”, “dataset” or more specific terms that make it clear what to make available; policies should include the benefits of data sharing; journals, publishers, and associations need to collaborate more to clarify data policies; and policies should explicitly ask for qualitative data. </p
Deep Learning for Histopathology Image Analysis From Heterogeneous and Multimodal Data Sources
No full textIn computational pathology, deep learning techniques have outperformed classical algorithms and hand-engineered features by leveraging large amounts of costly annotated data to train models that automatically learn relevant features for the tasks. Barriers to the successful application and design of deep learning approaches in computational pathology include: 1) The need for large image datasets with costly annotations to train deep learning models. 2) Visual variability of the images, which degrade the performance of models trained with homogeneous datasets. Finally, deep learning models in computational pathology usually leave out semantic information from pathology reports and other modalities that are complementary sources of information. This thesis contributes to solving these barriers by setting the following objectives: 1) To reduce the need for expensive pathologists’ annotations. 2) To evaluate and propose methods to overcome the visual heterogeneity of the images. 3) To exploit the semantic information from scientific literature and pathology reports
Same-same and different-different analogies through matching-to-sample tasks: A Pilot Study
No full textLa comprensión del lenguaje por medio del Análisis Experimental del Comportamiento ha sido un campo de estudio que se abordado por medio de la equivalencia de estímulos y la Teoría de los Marcos Relacionales haciendo uso de tareas de igualación a la muestra. De acuerdo con las pautas dadas por el experimento de Barnes y con los ajustes metodológicos de Ruiz, la presente investigación buscó realizar un pilotaje para identificar posibles errores en las fases de una tarea de razonamiento analógico y con ello reconocer factores disposicionales que puedan afectar el desempeño. Se encontró que existen diferencias en el número de aciertos entre las analogías similar-similar y diferente-diferente.Especialista en Psicología Clínica y Autoeficacia PersonalEspecializaciónThe understanding of language through Experimental Behavior Analysis has been a field of study that has been approached through the stimulus equivalence and the Relational Frame Theory using a matching to sample task. In accordance with the guidelines given by the Barnes experiment and with the methodological adjustments of Ruiz, the present investigation sought to carry out a pilot test to identify possible errors in the phases of an analog reasoning task and thereby recognize dispositional factors that may affect the throwput. It was found that there are differences in the number of hits between the similar-similar and different-different analogies
Deep learning-based retrieval system for gigapixel histopathology cases and the open access literature
No full textBackground: The introduction of digital pathology into clinical practice has led to the development of clinical workflows with digital images,
in connection with pathology reports. Still, most of the current work is time‑consuming manual analysis of image areas at different scales.
Links with data in the biomedical literature are rare, and a need for search based on visual similarity within whole slide images (WSIs) exists.
Objectives: The main objective of the work presented is to integrate content‑based visual retrieval with a WSI viewer in a prototype. Another
objective is to connect cases analyzed in the viewer with cases or images from the biomedical literature, including the search through visual
similarity and text. Methods: An innovative retrieval system for digital pathology is integrated with a WSI viewer, allowing to define regions of
interest (ROIs) in images as queries for finding visually similar areas in the same or other images and to zoom in/out to find structures at varying
magnification levels. The algorithms are based on a multimodal approach, exploiting both text information and content‑based image features.
Results: The retrieval system allows viewing WSIs and searching for regions that are visually similar to manually defined ROIs in various
data sources (proprietary and public datasets, e.g., scientific literature). The system was tested by pathologists, highlighting its capabilities and
suggesting ways to improve it and make it more usable in clinical practice. Conclusions: The developed system can enhance the practice of
pathologists by enabling them to use their experience and knowledge to control artificial intelligence tools for navigating repositories of images
for clinical decision support and teaching, where the comparison with visually similar cases can help to avoid misinterpretations. The system
is available as open source, allowing the scientific community to test, ideate and develop similar systems for research and clinical practice
- …
