Bidirectional Encoder Representations from Transformers (BERT) and BERT-based approaches are the current state-of-the-art in many natural language processing (NLP) tasks; however, their application to document classification on long clinical texts is limited. In this work, we introduce four methods scale BERT, which by default can only handle input sequences up approximately 400 words long, perform several thousand long. We compare these against two much simpler architectures - a word-level...

Abstract Objective We implement 2 different multitask learning (MTL) techniques, hard parameter sharing and cross-stitch, to train a word-level convolutional neural network (CNN) specifically designed for automatic extraction of cancer data from unstructured text in pathology reports. show the importance related information (IE) tasks leveraging shared representations across achieve state-of-the-art performance classification accuracy computational efficiency. Materials Methods Multitask CNN...

We introduce a deep learning architecture, hierarchical self-attention networks (HiSANs), designed for classifying pathology reports and show how its unique architecture leads to new state-of-the-art in accuracy, faster training, clear interpretability. evaluate performance on corpus of 374,899 obtained from the National Cancer Institute's (NCI) Surveillance, Epidemiology, End Results (SEER) program. Each report is associated with five clinical classification tasks - site, laterality,...

In real-world scenarios, ECG data are collected from a diverse range of heterogeneous devices, including high-end medical equipment and consumer-grade wearable each with varying computational capabilities constraints. This heterogeneity presents significant challenges in developing highly accurate deep learning (DL) global model for classification, as traditional centralized approaches struggle to address privacy concerns, scalability issues, inconsistencies arising device characteristics....

Population cancer registries can benefit from Deep Learning (DL) to automatically extract characteristics the high volume of unstructured pathology text reports they process annually. The success DL tackle this and other real-world problems is proportional availability large labeled datasets for model training. Although collaboration among essential fully exploit promise DL, privacy confidentiality concerns are main obstacles data sharing across registries. Moreover, natural language...

Background: Coronavirus disease 2019 (COVID-19), caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), emerged in Wuhan, China, late and created a global pandemic that overwhelmed healthcare systems. COVID-19, as of July 3, 2021, yielded 182 million confirmed cases 3.9 deaths globally according to the World Health Organization. Several patients who were initially diagnosed with mild or moderate COVID-19 later deteriorated reclassified type. Objective: The aim is create...

Automated text classification has many important applications in the clinical setting; however, obtaining labelled data for training machine learning and deep models is often difficult expensive. Active techniques may mitigate this challenge by reducing amount of required to effectively train a model. In study, we analyze effectiveness 11 active algorithms on classifying subsite histology from cancer pathology reports using Convolutional Neural Network as model.We compare performance each...

Effectively tackling the upcoming "zettabytes" data explosion requires a huge quantum leap in our computing power and energy efficiency. However, with Moore's law dwindling quickly, physical limits of CMOS technology make it almost intractable to achieve high efficiency if traditional "deterministic precise" model still dominates. Worse yet, mostly comprises statistics gleaned from uncertain, imperfect real-world environment. As such, means first principles modeling or explicit statistical...

Large-scale convolutional neural network is a fundamental algorithmic building block in many computer vision and artificial intelligence applications that follow the deep learning principle. However, typically-sized CNN well known to be computationally intensive. This work presents novel stochastic-based scalable hardware architecture circuit design computes large-scale with FPGA. The key idea implement all components of CNN, including multi-dimensional convolution, activation, pooling...

Information extraction and coding of free-text pathology reports is an important activity for cancer registries to support national surveillance. Cancer registrars must process high volumes on annual basis. In this study, we investigated automated approach using a coarse-to-fine training convolutional neural networks (CNNs) extracting the primary site, histological grade laterality from unstructured text reports. Our proposed scheme consists two stages. first stage, multi-task learning (MTL)...

Intrusion detection systems (IDS) are a very vital part of network security, as they can be used to protect the from illegal intrusions and communications. To detect malicious traffic, several IDS based on machine learning (ML) methods have been developed in literature. Machine models, other hand, recently proved effective, since vulnerable adversarial perturbations, which allows opponent crash system while performing queries. This motivated us present defensive model that uses training...

Individual electronic health records (EHRs) and clinical reports are often part of a larger sequence—for example, single patient may generate multiple over the trajectory disease. In applications such as cancer pathology reports, it is necessary not only to extract information from individual but also capture aggregate regarding entire case based off case-level context all in sequence. this paper, we introduce simple modular add-on for capturing that designed be compatible with most existing...

Cytogenetics laboratory tests are among the most important procedures for diagnosis of genetic diseases, especially in area hematological malignancies. Manual chromosomal karyotyping methods time consuming and labor intensive and, hence, expensive. Therefore, to alleviate process analysis, several attempts have been made enhance karyograms. The current image enhancement is based on classical processing. This approach has its limitations, one which that it a mandatory application all...

Automated text information extraction from cancer pathology reports is an active area of research to support national surveillance. A well-known challenge how develop tools with robust performance across registries. In this study we investigated whether transfer learning (TL) a convolutional neural network (CNN) can facilitate cross-registry knowledge sharing. Specifically, performed series experiments determine CNN trained single-registry data capable transferring another registry or...

High memory/storage complexity poses severe challenges to achieving high throughput and energy efficiency in discrete 2-D FIR filtering. This performance bottleneck is especially acute for embedded image or video applications, that use processing extensively, because real-time low power consumption are their paramount design objectives. Fortunately, most of such perception-based applications possess so-called "inherent fault tolerance", meaning slight computing accuracy degradation has a...

FPGA-based heterogeneous computing platform, due to its extreme logic reconfigurability, emerges be a strong contender as fabric in modern AI. As result, various accelerators for deep CNN—the key driver of AI—have been proposed their advantages high performance, and fast development round, etc. In general, the consensus among researchers is that, although accelerator can achieve much higher energy efficiency, raw performance lags behind when compared with GPUs similar density. this paper, we...

The rapid expansion and pervasive reach of the internet in recent years have raised concerns about evolving adaptable online threats, particularly with extensive integration Machine Learning (ML) systems into our daily routines. These are increasingly becoming targets malicious attacks that seek to distort their functionality through concept poisoning. Such aim warp intended operations these services, deviating them from true purpose. Poisoning renders susceptible unauthorized access,...

With large deep neural networks (DNNs) necessary to solve complex and data-intensive problems, energy efficiency is a key bottleneck for effectively deploying DL in the real world. Deep spiking NNs have gained much research attention recently due interest building biological availability of neuromorphic platforms, which can be orders magnitude more efficient compared CPUs GPUs. Although proven an technique solving many machine learning computer vision best our knowledge, this first attempt...

Deep learning has surged in popularity and proven to be effective for various artificial intelligence applications including information extraction from cancer pathology reports. Since word representation is a core unit that enables deep algorithms understand words able perform NLP, this must include as much possible help these achieve high classification performance. Therefore, work addition the distributional of large sized corpora, we use UMLS vocabulary resources enrich vector space with...

FIR filtering is widely used in many important DSP applications order to achieve stability and linear-phase property. This paper presents a hardware-and energy-efficient approach implement through reconfigurable stochastic computing. Specifically, we exploit basic probabilistic principle of summing independent random variables approximate without costly multiplications. allows our proposed architecture about 9 times 4 less power consumption than the conventional multiplier-based DA-based...

Energy efficiency and algorithmic robustness typically are conflicting circuit characteristics, yet with CMOS technology scaling towards 10-nm feature size, both become critical design metrics simultaneously for modern logic circuits. This paper propose a novel computing scheme hinged on probabilistic domain transformation aiming low power operation fault resilience. In such paradigm, algorithm inputs first encoded through means, which translates the input values into number of random...

Protein phosphorylation is a post-translational modification that enables various cellular activities and plays essential roles in protein interactions. Phosphorylation an important process for the replication of Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2). To shed more light on effects phosphorylation, we used ensemble neural networks to predict potential kinases might phosphorylate SARS-CoV-2 nonstructural proteins (nsps) molecular dynamics (MD) simulations investigate...

Finding optimal hyperparameters is necessary to identify the best performing deep learning models but process costly. In this paper, we applied model-based optimization, also known as Bayesian using CANDLE framework implemented on a High-Performance Computing environment. As use case selected information extraction from cancer pathology reports multi-task convolutional neural network, and hierarchical attention network be optimized. We utilized synthesized text corpus of 8,000 training cases...