The Tensor Brain: A Unified Theory of Perception, Memory, and Semantic Decoding.

We present a unified computational theory of an agent's perception and memory. In our model, both perception and memory are realized by different operational modes of the oscillating interactions between a symbolic index layer and a subsymbolic representation layer. The two layers form a bilayer tensor network (BTN). The index layer encodes indices for concepts, predicates, and episodic instances. The representation layer broadcasts information and reflects the cognitive brain state; it is our model of what authors have called the "mental canvas" or the "global workspace." As a bridge between perceptual input and the index layer, the representation layer enables the grounding of indices by their subsymbolic embeddings, which are implemented as connection weights linking both layers. The propagation of activation to earlier perceptual processing layers in the brain can lead to embodiments of indices. Perception and memories first create subsymbolic representations, which are subsequently decoded semantically to produce sequences of activated indices that form symbolic triple statements. The brain is a sampling engine: only activated indices are communicated to the remaining parts of the brain. Triple statements are dynamically embedded in the representation layer and embodied in earlier processing layers: the brain speaks to itself. Although memory appears to be about the past, its main purpose is to support the agent in the present and the future. Recent episodic memory provides the agent with a sense of the here and now. Remote episodic memory retrieves relevant past experiences to provide information about possible future scenarios. This aids the agent in decision making. "Future" episodic memory, based on expected future events, guides planning and action. Semantic memory retrieves specific information, which is not delivered by current perception, and defines priors for future observations. We argue that it is important for the agent to encode individual entities, not just classes and attributes. Perception is learning: episodic memories are constantly being formed, and we demonstrate that a form of self-supervised learning can acquire new concepts and refine existing ones. We test our model on a standard benchmark data set, which we expanded to contain richer representations for attributes, classes, and individuals. Our key hypothesis is that obtaining a better understanding of perception and memory is a crucial prerequisite to comprehending human-level intelligence.

Bringing Light Into the Dark: A Large-Scale Evaluation of Knowledge Graph Embedding Models Under a Unified Framework.

The heterogeneity in recently published knowledge graph embedding models' implementations, training, and evaluation has made fair and thorough comparisons difficult. To assess the reproducibility of previously published results, we re-implemented and evaluated 21 models in the PyKEEN software package. In this paper, we outline which results could be reproduced with their reported hyper-parameters, which could only be reproduced with alternate hyper-parameters, and which could not be reproduced at all, as well as provide insight as to why this might be the case. We then performed a large-scale benchmarking on four datasets with several thousands of experiments and 24,804 GPU hours of computation time. We present insights gained as to best practices, best configurations for each model, and where improvements could be made over previously published best configurations. Our results highlight that the combination of model architecture, training approach, loss function, and the explicit modeling of inverse relations is crucial for a model's performance and is not only determined by its architecture. We provide evidence that several architectures can obtain results competitive to the state of the art when configured carefully. We have made all code, experimental configurations, results, and analyses available at https://github.com/pykeen/pykeen and https://github.com/pykeen/benchmarking.

HUMAN-MACHINE COLLABORATION FOR MEDICAL IMAGE SEGMENTATION.

Image segmentation is a ubiquitous step in almost any medical image study. Deep learning-based approaches achieve state-of-the-art in the majority of image segmentation benchmarks. However, end-to-end training of such models requires sufficient annotation. In this paper, we propose a method based on conditional Generative Adversarial Network (cGAN) to address segmentation in semi-supervised setup and in a human-in-the-loop fashion. More specifically, we use the generator in the GAN to synthesize segmentations on unlabeled data and use the discriminator to identify unreliable slices for which expert annotation is required. The quantitative results on a conventional standard benchmark show that our method is comparable with the state-of-the-art fully supervised methods in slice-level evaluation, despite of requiring far less annotated data.

A highly accurate platform for clone-specific mutation discovery enables the study of active mutational processes.

Bulk whole genome sequencing (WGS) enables the analysis of tumor evolution but, because of depth limitations, can only identify old mutational events. The discovery of current mutational processes for predicting the tumor's evolutionary trajectory requires dense sequencing of individual clones or single cells. Such studies, however, are inherently problematic because of the discovery of excessive false positive (FP) mutations when sequencing picogram quantities of DNA. Data pooling to increase the confidence in the discovered mutations, moves the discovery back in the past to a common ancestor. Here we report a robust WGS and analysis pipeline (DigiPico/MutLX) that virtually eliminates all F results while retaining an excellent proportion of true positives. Using our method, we identified, for the first time, a hyper-mutation (kataegis) event in a group of ∼30 cancer cells from a recurrent ovarian carcinoma. This was unidentifiable from the bulk WGS data. Overall, we propose DigiPico/MutLX method as a powerful framework for the identification of clone-specific variants at an unprecedented accuracy.

Predicting Visual Acuity by Using Machine Learning in Patients Treated for Neovascular Age-Related Macular Degeneration.

PURPOSE: To predict, by using machine learning, visual acuity (VA) at 3 and 12 months in patients with neovascular age-related macular degeneration (AMD) after initial upload of 3 anti-vascular endothelial growth factor (VEGF) injections. DESIGN: Database study. PARTICIPANTS: For the 3-month VA forecast, 653 patients (379 female) with 738 eyes and an average age of 74.1 years were included. The baseline VA before the first injection was 0.54 logarithm of the minimum angle of resolution (logMAR) (±0.39). A total of 456 of these patients (270 female, 508 eyes, average age: 74.2 years) had sufficient follow-up data to be included for a 12-month VA prediction. The baseline VA before the first injection was 0.56 logMAR (±0.42). METHODS: Five different machine-learning algorithms (AdaBoost.R2, Gradient Boosting, Random Forests, Extremely Randomized Trees, and Lasso) were used to predict VA in patients with neovascular AMD after treatment with 3 anti-VEGF injections. Clinical data features came from a data warehouse (DW) containing electronic medical records (41 features, e.g., VA) and measurement features from OCT (124 features, e.g., central retinal thickness). The VA of patient eyes excluded from machine learning was predicted and compared with the ground truth, namely, the actual VA of these patients as recorded in the DW. MAIN OUTCOME MEASURES: Difference in logMAR VA after 3 and 12 months upload phase between prediction and ground truth as defined. RESULTS: For the 3-month VA forecast, the difference between the prediction and ground truth was between 0.11 logMAR (5.5 letters) mean absolute error (MAE)/0.14 logMAR (7 letters) root mean square error (RMSE) and 0.18 logMAR (9 letters) MAE/0.2 logMAR (10 letters) RMSE. For the 12-month VA forecast, the difference between the prediction and ground truth was between 0.16 logMAR (8 letters) MAE/0.2 logMAR (10 letters) RMSE and 0.22 logMAR (11 letters) MAE/0.26 logMAR (13 letters) RMSE. The best performing algorithm was the Lasso protocol. CONCLUSIONS: Machine learning allowed VA to be predicted for 3 months with a comparable result to VA measurement reliability. For a forecast after 12 months of therapy, VA prediction may help to encourage patients adhering to intravitreal therapy.

Extraction of semantic biomedical relations from text using conditional random fields.

BACKGROUND: The increasing amount of published literature in biomedicine represents an immense source of knowledge, which can only efficiently be accessed by a new generation of automated information extraction tools. Named entity recognition of well-defined objects, such as genes or proteins, has achieved a sufficient level of maturity such that it can form the basis for the next step: the extraction of relations that exist between the recognized entities. Whereas most early work focused on the mere detection of relations, the classification of the type of relation is also of great importance and this is the focus of this work. In this paper we describe an approach that extracts both the existence of a relation and its type. Our work is based on Conditional Random Fields, which have been applied with much success to the task of named entity recognition. RESULTS: We benchmark our approach on two different tasks. The first task is the identification of semantic relations between diseases and treatments. The available data set consists of manually annotated PubMed abstracts. The second task is the identification of relations between genes and diseases from a set of concise phrases, so-called GeneRIF (Gene Reference Into Function) phrases. In our experimental setting, we do not assume that the entities are given, as is often the case in previous relation extraction work. Rather the extraction of the entities is solved as a subproblem. Compared with other state-of-the-art approaches, we achieve very competitive results on both data sets. To demonstrate the scalability of our solution, we apply our approach to the complete human GeneRIF database. The resulting gene-disease network contains 34758 semantic associations between 4939 genes and 1745 diseases. The gene-disease network is publicly available as a machine-readable RDF graph. CONCLUSION: We extend the framework of Conditional Random Fields towards the annotation of semantic relations from text and apply it to the biomedical domain. Our approach is based on a rich set of textual features and achieves a performance that is competitive to leading approaches. The model is quite general and can be extended to handle arbitrary biological entities and relation types. The resulting gene-disease network shows that the GeneRIF database provides a rich knowledge source for text mining. Current work is focused on improving the accuracy of detection of entities as well as entity boundaries, which will also greatly improve the relation extraction performance.

Mining functional modules in genetic networks with decomposable graphical models.

In recent years, graphical models have become an increasingly important tool for the structural analysis of genome-wide expression profiles at the systems level. Here we present a new graphical modelling technique, which is based on decomposable graphical models, and apply it to a set of gene expression profiles from acute lymphoblastic leukemia (ALL). The new method explains probabilistic dependencies of expression levels in terms of the concerted action of underlying genetic functional modules, which are represented as so-called "cliques" in the graph. In addition, the method uses continuous-valued (instead of discretized) expression levels, and makes no particular assumption about their probability distribution. We show that the method successfully groups members of known functional modules to cliques. Our method allows the evaluation of the importance of genes for global cellular functions based on both link count and the clique membership count.

Classification of rheumatoid joint inflammation based on laser imaging.

We describe a classification system for a novel imaging method for arthritic finger joints. The basis of this system is a laser imaging technique which is sensitive to the optical characteristics of finger joint tissue. From the laser images acquired at baseline and follow-up, finger joints can automatically be classified according to whether the inflammatory status has improved or worsened. To perform the classification task, various linear and kernel-based systems were implemented and their performances were compared. Based on the results presented in this paper, we conclude that the laser-based imaging permits a reliable classification of pathological finger joints, making it a sensitive method for detecting arthritic changes.

Assessment of proximal finger joint inflammation in patients with rheumatoid arthritis, using a novel laser-based imaging technique.

OBJECTIVE: To evaluate a newly developed laser-based imaging technique for the study of soft tissue changes and acute inflammatory processes of the proximal interphalangeal (PIP) joints in patients with rheumatoid arthritis (RA). METHODS: A novel imaging device was developed which allows the transillumination of PIP joints using laser light in the near-infrared wavelength range. In a first clinical followup study, a total of 72 PIP joints of 22 patients with RA and 64 PIP joints of 8 healthy controls were examined both clinically and with the new laser device. At baseline and at followup after a mean of 6 weeks, clinical signs of synovitis, the joint circumference, and the degree of pain were assessed for each PIP joint in order to determine the clinical degree of inflammation. Different features were extracted from the laser images and evaluated by a neural network. RESULTS: At baseline, 72 PIP joints in the RA patients showed clinical signs of inflammation. At followup, 45 PIP joints showed clinical improvement, 13 showed steady active inflammation, and 14 showed deterioration compared with the first visit. None of the 64 PIP joints in the healthy individuals showed any signs of synovitis. The inflammatory status of 60 of the 72 RA joints examined was classified correctly by laser examination and joint circumference determination, giving a sensitivity of 80%, a specificity of 89%, and an accuracy of 83% in detecting inflammatory changes in affected joints. Laser data and joint circumference determination of healthy joints at followup resulted in an accuracy of 85% in reproducing the image. CONCLUSION: The new laser-based imaging technique allows the transillumination of PIP joints and gives information about the inflammatory status of the joint after processing through a neural network. Our data indicate that laser imaging may provide additional information in the early diagnosis of an inflammatory joint process and may prove particularly useful in assessing acute joint inflammation at followup.