Bayesian Estimate of Mean Proper Scores for Diversity-Enhanced Active Learning

The effectiveness of active learning largely depends on the sampling efficiency of the acquisition function. Expected Loss Reduction (ELR) focuses on a Bayesian estimate of the reduction in classification error, and more general costs fit in the same framework. We propose Bayesian Estimate of Mean Proper Scores (BEMPS) to estimate the increase in strictly proper scores such as log probability or negative mean square error within this framework. We also prove convergence results for this general class of costs. To facilitate better experimentation with the new acquisition functions, we develop a complementary batch AL algorithm that encourages diversity in the vector of expected changes in scores for unlabeled data. To allow high-performance classifiers, we combine deep ensembles, and dynamic validation set construction on pretrained models, and further speed up the ensemble process with the idea of Monte Carlo Dropout. Extensive experiments on both texts and images show that the use of mean square error and log probability with BEMPS yields robust acquisition functions and well-calibrated classifiers, and consistently outperforms the others tested. The advantages of BEMPS over the others are further supported by a set of qualitative analyses, where we visualise their sampling behaviour using data maps and t-SNE plots.

A BERT model generates diagnostically relevant semantic embeddings from pathology synopses with active learning

Background

Pathology synopses consist of semi-structured or unstructured text summarizing visual information by observing human tissue. Experts write and interpret these synopses with high domain-specific knowledge to extract tissue semantics and formulate a diagnosis in the context of ancillary testing and clinical information. The limited number of specialists available to interpret pathology synopses restricts the utility of the inherent information. Deep learning offers a tool for information extraction and automatic feature generation from complex datasets.

Methods

Using an active learning approach, we developed a set of semantic labels for bone marrow aspirate pathology synopses. We then trained a transformer-based deep-learning model to map these synopses to one or more semantic labels, and extracted learned embeddings (i.e., meaningful attributes) from the model's hidden layer.

Results

Here we demonstrate that with a small amount of training data, a transformer-based natural language model can extract embeddings from pathology synopses that capture diagnostically relevant information. On average, these embeddings can be used to generate semantic labels mapping patients to probable diagnostic groups with a micro-average F1 score of 0.779 Â ± 0.025.

Conclusions

We provide a generalizable deep learning model and approach to unlock the semantic information inherent in pathology synopses toward improved diagnostics, biodiscovery and AI-assisted computational pathology.

Intelligent Labeling Based on Fisher Information for Medical Image Segmentation Using Deep Learning

Deep convolutional neural networks (CNN) have recently achieved superior performance at the task of medical image segmentation compared to classic models. However, training a generalizable CNN requires a large amount of training data, which is difficult, expensive, and time-consuming to obtain in medical settings. Active Learning (AL) algorithms can facilitate training CNN models by proposing a small number of the most informative data samples to be annotated to achieve a rapid increase in performance. We proposed a new active learning method based on Fisher information (FI) for CNNs for the first time. Using efficient backpropagation methods for computing gradients together with a novel low-dimensional approximation of FI enabled us to compute FI for CNNs with a large number of parameters. We evaluated the proposed method for brain extraction with a patch-wise segmentation CNN model in two different learning scenarios: universal active learning and active semi-automatic segmentation. In both scenarios, an initial model was obtained using labeled training subjects of a source data set and the goal was to annotate a small subset of new samples to build a model that performs well on the target subject(s). The target data sets included images that differed from the source data by either age group (e.g. newborns with different image contrast) or underlying pathology that was not available in the source data. In comparison to several recently proposed AL methods and brain extraction baselines, the results showed that FI-based AL outperformed the competing methods in improving the performance of the model after labeling a very small portion of target data set (<0.25%).

Active 3-D Shape Cosegmentation With Graph Convolutional Networks

We present a novel active learning approach for shape cosegmentation based on graph convolutional networks (GCNs). The premise of our approach is to represent the collections of three-dimensional shapes as graph-structured data, where each node in the graph corresponds to a primitive patch of an oversegmented shape, and is associated with a representation initialized by extracting features. Then, the GCN operates directly on the graph to update the representation of each node based on a layer-wise propagation rule, which aggregates information from its neighbors, and predicts the labels for unlabeled nodes. Additionally, we further suggest an active learning strategy that queries the most informative samples to extend the initial training samples of GCN to generate more accurate predictions of our method. Our experimental results on the Shape COSEG dataset demonstrate the effectiveness of our approach.

A Semi-Automatic Annotation Approach for Human Activity Recognition

Modern smartphones and wearables often contain multiple embedded sensors which generate significant amounts of data. This information can be used for body monitoring-based areas such as healthcare, indoor location, user-adaptive recommendations and transportation. The development of Human Activity Recognition (HAR) algorithms involves the collection of a large amount of labelled data which should be annotated by an expert. However, the data annotation process on large datasets is expensive, time consuming and difficult to obtain. The development of a HAR approach which requires low annotation effort and still maintains adequate performance is a relevant challenge. We introduce a Semi-Supervised Active Learning (SSAL) based on Self-Training (ST) approach for Human Activity Recognition to partially automate the annotation process, reducing the annotation effort and the required volume of annotated data to obtain a high performance classifier. Our approach uses a criterion to select the most relevant samples for annotation by the expert and propagate their label to the most confident samples. We present a comprehensive study comparing supervised and unsupervised methods with our approach on two datasets composed of daily living activities. The results showed that it is possible to reduce the required annotated data by more than 89% while still maintaining an accurate model performance.