Fine-tuned Neural Network Research Articles

Skin Cancer Classification Using Fine-Tuned Transfer Learning of DENSENET-121

Skin cancer diagnosis greatly benefits from advanced machine learning techniques, particularly fine-tuned deep learning models. In our research, we explored the impact of traditional machine learning and fine-tuned deep learning approaches on prediction accuracy. Our findings reveal significant improvements in predictability and accuracy with fine-tuning, particularly evident in deep learning models. The CNN, SVM, and Random Forest Classifier achieved high accuracy. However, fine-tuned deep learning models such as EfficientNetB0, ResNet34, VGG16, Inception _v3, and DenseNet121 demonstrated superior performance. To ensure comparability, we fine-tuned these models by incorporating additional layers, including one flatten layer and three densely interconnected layers. These layers play a crucial role in enhancing model efficiency and performance. The flatten layer preprocesses multidimensional feature maps, facilitating efficient information flow, while subsequent dense layers refine feature representations, capturing intricate patterns and relationships within the data. Leveraging LeakyReLU activation functions in the dense layers mitigates the vanishing gradient problem and promotes stable training. Finally, the output dense layer with a sigmoid activation function simplifies decision making for healthcare professionals by providing binary classification output. Our study underscores the significance of incorporating additional layers in fine-tuned neural network models for skin cancer classification, offering improved accuracy and reliability in diagnosis.

Neural Network Model for Predicting Progression of Disease

This study investigates the use of neural network and their ability to predict disease progression based on clinical data and biomarkers. Using deep neural networks, a model was developed that efficiently analyzes the complex relationship between various factors and predict the probability of disease. The model was validated using retrospective analysis which indicated a good predictive ability that could be further utilized in better diagnostics and personalized treatment methods. More importantly, reserch detected specific pattern in the data, which enabled a more accurate prediction of disease at different stages. The study tried to improve a model by fine-tuned neural networks and tested other frameworks to gain the highets precision. This research also provides a basic for future work in directing the development of personalized therapeutic approaches based on individual patient characteristics.

Robust diagnosis and meta visualizations of plant diseases through deep neural architecture with explainable AI

Deep learning has emerged as a highly effective and precise method for classifying images. The presence of plant diseases poses a significant threat to food security. However, accurately identifying these diseases in plants is challenging due to limited infrastructure and techniques. Fortunately, the recent advancements in deep learning within the field of computer vision have opened up new possibilities for diagnosing plant pathology. Detecting plant diseases at an early stage is crucial, and this research paper proposes a deep convolutional neural network model that can rapidly and accurately identify plant diseases. Given the minimal variation in image texture and color, deep learning techniques are essential for robust recognition. In this study, we introduce a deep, explainable neural architecture specifically designed for recognizing plant diseases. Fine-tuned deep convolutional neural network is designed by freezing the layers and adjusting the weights of learnable layers. By extracting deep features from a down sampled feature map of a fine-tuned neural network, we are able to classify these features using a customized K-Nearest Neighbors Algorithm. To train and validate our model, we utilize the largest standard plant village dataset, which consists of 38 classes. To evaluate the performance of our proposed system, we estimate specificity, sensitivity, accuracy, and AUC. The results demonstrate that our system achieves an impressive maximum validation accuracy of 99.95% and an AUC of 1, making it the most ideal and highest-performing approach compared to current state-of-the-art deep learning methods for automatically identifying plant diseases.

Deep-Learning-Guided Evaluation Method for the High-Volume Preparation of Flexible Sensors Based on Inkjet Printing.

Flexible sensors for application in various industries, including biomedicine and wearable electronics, are frequently made using silver nanoparticle (AgNP) inks and inkjet printing (IJP) technology. Inkjet-printed flexible electronic devices are made up of many printed lines that run parallel to each other, and the surface morphology of the printed lines and the interline state directly impact the electrical conductivity of the electronic devices. This paper describes the experimental setup for IJP, the definition of print line characteristics, and common unavoidable defects. Conductivity and physical defects are considered in defining the print line quality assessment. In addition, two prediction models of flexible sensors before batch printing and a model for detecting defects after printing are provided. The predictive models can guide actions, leading to a print success rate of over 80%. We build the defect detection model using a neural architecture search because manually fine-tuning neural networks for reference is challenging. Finally, a target detection model with a mAP@0.5 of 81.2% is built in just 0.77 graphics processing unit (GPU) days. The model takes only 4.6 ms to detect an image, satisfying the real-time monitoring needs. At the same time, an accuracy of 95.5% can be achieved in the test data set. This work provides a new idea for the high-volume preparation of flexible sensors.

A New Method for Integrating a Deep Learning Module into Special Object Classification Software

The article discusses a method for integrating a deep learning module into special software based on classical computer vision methods. This software implements the functions of creating panoramic images and their subsequent comparison with the ability to highlight different areas. The integration of deep learning methods is aimed at automating the classification of found objects and solving problems associated with identifying a large number of small objects and noise anomalies. In the process of carrying out the research, an additional neural network generative model was created that improves the quality of classification by increasing the size of the training dataset. This model was implemented in the form of an ensemble of variational autoencoders, which made it possible to simplify the learning process and fine-tuning of neural networks. Also, in order to assess the quality of the ensemble of variational autoencoders, a new method for checking the generated synthetic data was developed. The result of the integration was improved special software capable of classifying the detected found objects of interest with high accuracy.

Early detection of hepatocellular carcinoma via no end-repair enzymatic methylation sequencing of cell-free DNA and pre-trained neural network

BackgroundEarly detection of hepatocellular carcinoma (HCC) is important in order to improve patient prognosis and survival rate. Methylation sequencing combined with neural networks to identify cell-free DNA (cfDNA) carrying aberrant methylation offers an appealing and non-invasive approach for HCC detection. However, some limitations exist in traditional methylation detection technologies and models, which may impede their performance in the read-level detection of HCC.MethodsWe developed a low DNA damage and high-fidelity methylation detection method called No End-repair Enzymatic Methyl-seq (NEEM-seq). We further developed a read-level neural detection model called DeepTrace that can better identify HCC-derived sequencing reads through a pre-trained and fine-tuned neural network. After pre-training on 11 million reads from NEEM-seq, DeepTrace was fine-tuned using 1.2 million HCC-derived reads from tumor tissue DNA after noise reduction, and 2.7 million non-tumor reads from non-tumor cfDNA. We validated the model using data from 130 individuals with cfDNA whole-genome NEEM-seq at around 1.6X depth.ResultsNEEM-seq overcomes the drawbacks of traditional enzymatic methylation sequencing methods by avoiding the introduction of unmethylation errors in cfDNA. DeepTrace outperformed other models in identifying HCC-derived reads and detecting HCC individuals. Based on the whole-genome NEEM-seq data of cfDNA, our model showed high accuracy of 96.2%, sensitivity of 93.6%, and specificity of 98.5% in the validation cohort consisting of 62 HCC patients, 48 liver disease patients, and 20 healthy individuals. In the early stage of HCC (BCLC 0/A and TNM I), the sensitivity of DeepTrace was 89.6 and 89.5% respectively, outperforming Alpha Fetoprotein (AFP) which showed much lower sensitivity in both BCLC 0/A (50.5%) and TNM I (44.7%).ConclusionsBy combining high-fidelity methylation data from NEEM-seq with the DeepTrace model, our method has great potential for HCC early detection with high sensitivity and specificity, making it potentially suitable for clinical applications.DeepTrace: https://github.com/Bamrock/DeepTrace

Measuring Job Loss during the Pandemic Recession in Real Time with Twitter Data

We present an indicator of job loss derived from Twitter data, based on a fine-tuned neural network with transfer learning to classify if a tweet is job-loss related or not. We show that our Twitter-based measure of job loss is well-correlated with and predictive of other measures of unemployment available in the official statistics and with the added benefits of real-time availability and daily frequency. These findings are especially strong for the period of the Pandemic Recession, when our Twitter indicator continues to track job loss well but where other real-time measures like unemployment insurance claims provided an imperfect signal of job loss. Additionally, we find that our Twitter job loss indicator provides incremental information in predicting official unemployment flows in a given month beyond what weekly unemployment insurance claims offer.

Geo-Spatial Mapping of Hate Speech Prediction in Roman Urdu

Social media has transformed into a crucial channel for political expression. Twitter, especially, is a vital platform used to exchange political hate in Pakistan. Political hate speech affects the public image of politicians, targets their supporters, and hurts public sentiments. Hate speech is a controversial public speech that promotes violence toward a person or group based on specific characteristics. Although studies have been conducted to identify hate speech in European languages, Roman languages have yet to receive much attention. In this research work, we present the automatic detection of political hate speech in Roman Urdu. An exclusive political hate speech labeled dataset (RU-PHS) containing 5002 instances and city-level information has been developed. To overcome the vast lexical structure of Roman Urdu, we propose an algorithm for the lexical unification of Roman Urdu. Three vectorization techniques are developed: TF-IDF, word2vec, and fastText. A comparative analysis of the accuracy and time complexity of conventional machine learning models and fine-tuned neural networks using dense word representations is presented for classifying and predicting political hate speech. The results show that a random forest and the proposed feed-forward neural network achieve an accuracy of 93% using fastText word embedding to distinguish between neutral and politically offensive speech. The statistical information helps identify trends and patterns, and the hotspot and cluster analysis assist in pinpointing Punjab as a highly susceptible area in Pakistan in terms of political hate tweet generation.

The context effect for blind image quality assessment

Image quality assessment (IQA) is a process of visuo-cognitive, which is an essential stage in human interaction with the environment. The study of the context effect (Brown and Daniel, 1987) also shows that the evaluation results made by the human vision system (HVS) is related to the contrast between the distorted image and the background environment. However, the existing IQA methods carry out the quality evaluation that only depends on the distorted image itself and ignores the impact of environment to human perception. In this paper, we propose a novel blind image quality assessment(BIQA) based on the context effect. At first, we use a graphical model to describe how the context effect influences human perception of image quality. Based on the established graph, we construct the context relation between the distorted image and the background environment by the X. Han et al. (2015). Then the context features are extracted from the constructed relation, and the quality-related features are extracted by the fine-tuned neural network from the distorted image in pixel-wise. Finally, these features are concatenated to quantify image quality degradations and then regress to quality scores. In addition, the proposed method is adaptive to various deep neural networks. Experimental results show that the proposed method not only has the state-of-art performance on the synthetic distorted images, but also has a great improvement on the authentic distorted images.

Clinical Trial Classification of SNS24 Calls with Neural Networks

SNS24, the Portuguese National Health Contact Center, is a telephone and digital public service that provides clinical services. SNS24 plays an important role in the identification of users’ clinical situations according to their symptoms. Currently, there are a number of possible clinical algorithms defined, and selecting the appropriate clinical algorithm is very important in each telephone triage episode. Decreasing the duration of the phone calls and allowing a faster interaction between citizens and SNS24 service can further improve the performance of the telephone triage service. In this paper, we present a study using deep learning approaches to build classification models, aiming to support the nurses with the clinical algorithm’s choice. Three different deep learning architectures, namely convolutional neural network (CNN), recurrent neural network (RNN), and transformers-based approaches are applied across a total number of 269,654 call records belonging to 51 classes. The CNN, RNN, and transformers-based model each achieve an accuracy of 76.56%, 75.88%, and 78.15% over the test set in the preliminary experiments. Models using the transformers-based architecture are further fine-tuned, achieving an accuracy of 79.67% with Adam and 79.72% with SGD after learning rate fine-tuning; an accuracy of 79.96% with Adam and 79.76% with SGD after epochs fine-tuning; an accuracy of 80.57% with Adam after the batch size fine-tuning. Analysis of similar clinical symptoms is carried out using the fine-tuned neural network model. Comparisons are done over the labels predicted by the neural network model, the support vector machines model, and the original labels from SNS24. These results suggest that using deep learning is an effective and promising approach to aid the clinical triage of the SNS24 phone call services.

Real-world Patient Trajectory Prediction from Clinical Notes Using Artificial Neural Networks and UMLS-Based Extraction of Concepts.

As more data is generated from medical attendances and as Artificial Neural Networks gain momentum in research and industry, computer-aided medical prognosis has become a promising technology. A common approach to perform automated prognoses relies on textual clinical notes extracted from Electronic Health Records (EHRs). Data from EHRs are fed to neural networks that produce a set with the most probable medical problems to which a patient is subject in her/his clinical future, including clinical conditions, mortality, and readmission. Following this research line, we introduce a methodology that takes advantage of the unstructured text found in clinical notes by applying preprocessing, concepts extraction, and fine-tuned neural networks to predict the most probable medical problems to follow in a patient's clinical trajectory. Different from former works that focus on word embeddings and raw sets of extracted concepts, we generate a refined set of Unified Medical Language System (UMLS) concepts by applying a similarity threshold filter and a list of acceptable concept types. In our prediction experiments, our method demonstrated AUC-ROC performance of 0.91 for diagnosis codes, 0.93 for mortality, and 0.72 for readmission, determining an efficacy that rivals state-of-the-art works. Our findings contribute to the development of automated prognosis systems in hospitals where text is the main source of clinical history.

Performance of a U-Net-based neural network for predictive adaptive optics.

We apply a U-Net-based convolutional neural network (NN) architecture to the problem of predictive adaptive optics (AO) for tracking and imaging fast-moving targets, such as satellites in low Earth orbit (LEO). We show that the fine-tuned NN is able to achieve an approximately 50% reduction in mean-squared wavefront error over non-predictive approaches while predicting up to eight frames into the future. These results were obtained when the NN, trained mostly on simulated data, tested its performance on 1 kHz Shack-Hartmann wavefront sensor data collected in open-loop at the Advanced Electro-Optical System facility at Haleakala Observatory while the telescope tracked a naturally illuminated piece of LEO space debris. We report, to our knowledge, the first successful test of a NN for the predictive AO application using on-sky data, as well as the first time such a network has been developed for the more stressing space tracking application.

Classification of Skin Disease Using Deep Learning Neural Networks with MobileNet V2 and LSTM.

Deep learning models are efficient in learning the features that assist in understanding complex patterns precisely. This study proposed a computerized process of classifying skin disease through deep learning based MobileNet V2 and Long Short Term Memory (LSTM). The MobileNet V2 model proved to be efficient with a better accuracy that can work on lightweight computational devices. The proposed model is efficient in maintaining stateful information for precise predictions. A grey-level co-occurrence matrix is used for assessing the progress of diseased growth. The performance has been compared against other state-of-the-art models such as Fine-Tuned Neural Networks (FTNN), Convolutional Neural Network (CNN), Very Deep Convolutional Networks for Large-Scale Image Recognition developed by Visual Geometry Group (VGG), and convolutional neural network architecture that expanded with few changes. The HAM10000 dataset is used and the proposed method has outperformed other methods with more than 85% accuracy. Its robustness in recognizing the affected region much faster with almost 2× lesser computations than the conventional MobileNet model results in minimal computational efforts. Furthermore, a mobile application is designed for instant and proper action. It helps the patient and dermatologists identify the type of disease from the affected region’s image at the initial stage of the skin disease. These findings suggest that the proposed system can help general practitioners efficiently and effectively diagnose skin conditions, thereby reducing further complications and morbidity.

ACTL: Asymmetric Convolutional Transfer Learning for Tree Species Identification Based on Deep Neural Network

The identification of tree species is of great significance to the sustainable management and utilization of forest ecosystems. Hyperspectral data provide sufficient spectral and spatial information to classify tree species. Convolutional neural networks (CNN) have achieved great success in hyperspectral image (HSI) classification. The outstanding performance of CNN in HSI classification relies on sufficient training samples. However, it’s expensive and time consuming to acquire labeled training samples. In this article, a novel asymmetric convolutional transfer learning model for HSI classification is proposed. First, the tree species identification dataset is built from Goddard’s LiDAR, Hyperspectral & Thermal (G-LiHT) data. Then, the asymmetric convolutional transfer learning model and weights trained on ImageNet dataset are used to initialize the weights of the HSI classification model. Finally, a well fine-tuned neural network on tree species dataset is used to perform the HSI classification task. The experimental results reveal that the proposed model with asymmetric convolutional blocks effectively improves the accuracy of Howland forest tree species identification and provides a new idea for the classification of hyperspectral remote sensing images.

Aerial Scene Classification through Fine-Tuning with Adaptive Learning Rates and Label Smoothing

Remote Sensing (RS) image classification has recently attracted great attention for its application in different tasks, including environmental monitoring, battlefield surveillance, and geospatial object detection. The best practices for these tasks often involve transfer learning from pre-trained Convolutional Neural Networks (CNNs). A common approach in the literature is employing CNNs for feature extraction, and subsequently train classifiers exploiting such features. In this paper, we propose the adoption of transfer learning by fine-tuning pre-trained CNNs for end-to-end aerial image classification. Our approach performs feature extraction from the fine-tuned neural networks and remote sensing image classification with a Support Vector Machine (SVM) model with linear and Radial Basis Function (RBF) kernels. To tune the learning rate hyperparameter, we employ a linear decay learning rate scheduler as well as cyclical learning rates. Moreover, in order to mitigate the overfitting problem of pre-trained models, we apply label smoothing regularization. For the fine-tuning and feature extraction process, we adopt the Inception-v3 and Xception inception-based CNNs, as well the residual-based networks ResNet50 and DenseNet121. We present extensive experiments on two real-world remote sensing image datasets: AID and NWPU-RESISC45. The results show that the proposed method exhibits classification accuracy of up to 98%, outperforming other state-of-the-art methods.

PS02.070: DEEP NEURAL NETWORK TO PREDICT POOR PROGNOSTIC FACTORS IN PATIENTS WITH ESOPHAGEAL CANCER

Abstract Background Both lymphovascular invasion, which is characterized by penetration of tumor cells into the peritumoural vascular or lymphatic network, and perineural invasion, which is characterized by involvement of tumor cells surrounding nerve fibers, are considered as an important step for tumor spreading, and are known poor prognostic factors in esophageal cancer. However, the information of these histological features is unavailable until pathological examination of surgical resected specimens. We aim to predict the presence or absence of these factors by positron emission tomography images during staging workup. Methods The positron emission tomography images before treatment and pathological reports of 278 patients who underwent esophagectomy for squamous cell carcinoma were collected. Stepwise convolutional neural network was constructed to distinguish patient with either lymphovascular invasion or perineural invasion from those without. Results Randomly selected 248 patients were included in the testing set. Stepwise approach was used in training our custom neural network. The performance of fine-tuned neural network was tested in another independent 30 patients. The accuracy rate of predicting the presence or absence of either lymphovascular invasion or perineural invasion was 66.7% (20 of 30 were accurate). Conclusion Using pre-treatment positron emission tomography images alone to predict the presence of absence of poor prognostic histological factors, i.e. lymphovascular invasion or perineural invasion, with deep convolutional neural network is possible. The technique of deep learning may identify patients with poor prognosis and enable personalized medicine in esophageal cancer. Disclosure All authors have declared no conflicts of interest.

Conceptor-based deep neural networks

In recent years, deep neural networks, also known as deep learning, have achieved several breakthroughs in different fields that were previously dominated by machine learning. Even when using high-performance computing devices, it takes days or weeks to train a deep neural network. Conceptor, as an extension of echo state networks, can be understood as certain neural filters that characterize dynamical neural activation patterns. In this study, based on some improvements to the original conceptor model, we have conducted several studies from the perspectives of non-iterative methods and transfer learning to address the issues mentioned above, which can be summarized as follows: (1) A conceptor-based classifier for non-temporal data and a non-iterative approach feedforward convolutional conceptor neural network are proposed. This classifier achieves classifying accuracy comparable to that of the state-of-the-art methods while requiring significantly less training time. Through experiments on MNIST variation datasets, we evaluate the classifying quality of the feedforward convolutional conceptor neural network. (2) A classifier called fast conceptor classifier is proposed based on conceptors and it achieves state-of-the-art results with the training time reduced by a factor of 60 on average. Its evaluations with pre-trained rather than fine-tuned neural networks have been investigated on Caltech-101 and Caltech-256 datasets.

Single-label and multi-label conceptor classifiers in pre-trained neural networks

Training large neural network models from scratch is not feasible due to over-fitting on small datasets and too much time consumed on large datasets. To address this, transfer learning, namely utilizing the feature extracting capacity learned by large models, becomes a hot spot in neural network community. At the classifying stage of pre-trained neural network model, either a linear SVM classifier or a Softmax classifier is employed and that is the only trained part of the whole model. In this paper, inspired by transfer learning, we propose a classifier based on conceptors called Multi-label Conceptor Classifier (MCC) to deal with multi-label classification in pre-trained neural networks. When no multi-label sample exists, MCC equates to Fast Conceptor Classifier, a fast single-label classifier proposed in our previous work, thus being applicable to single-label classification. Moreover, by introducing a random search algorithm, we further improve the performance of MCC on single-label datasets Caltech-101 and Caltech-256, where it achieves state-of-the-art results. Also, its evaluations with pre-trained rather than fine-tuning neural networks are investigated on multi-label dataset PASCAL VOC-2007, where it achieves comparable results.

Family of FLP Peptides in Caenorhabditis elegans and Related Nematodes.

Neuropeptides regulate all aspects of behavior in multicellular organisms. Because of their ability to act at long distances, neuropeptides can exert their effects beyond the conventional synaptic connections, thereby adding an intricate layer of complexity to the activity of neural networks. In the nematode Caenorhabditis elegans, a large number of neuropeptide genes that are expressed throughout the nervous system have been identified. The actions of these peptides supplement the synaptic connections of the 302 neurons, allowing for fine tuning of neural networks and increasing the ways in which behaviors can be regulated. In this review, we focus on a large family of genes encoding FMRFamide-related peptides (FaRPs). These genes, the flp genes, have been used as a starting point to identifying flp genes throughout Nematoda. Nematodes have the largest family of FaRPs described thus far. The challenges in the future are the elucidation of their functions and the identification of the receptors and signaling pathways through which they function.

The BDNF effects on dendritic spines of mature hippocampal neurons depend on neuronal activity

The fine tuning of neural networks during development and learning relies upon both functional and structural plastic processes. Changes in the number as well as in the size and shape of dendritic spines are associated to long-term activity-dependent synaptic plasticity. However, the molecular mechanisms translating functional into structural changes are still largely unknown. In this context, neurotrophins, like Brain-Derived Neurotrophic Factor (BDNF), are among promising candidates. Specifically BDNF-TrkB receptor signaling is crucial for activity-dependent strengthening of synapses in different brain regions. BDNF application has been shown to positively modulate dendritic and spine architecture in cortical and hippocampal neurons as well as structural plasticity in vitro. However, a global BDNF deprivation throughout the central nervous system (CNS) resulted in very mild structural alterations of dendritic spines, questioning the relevance of the endogenous BDNF signaling in modulating the development and the mature structure of neurons in vivo. Here we show that a loss-of-function approach, blocking BDNF results in a significant reduction in dendritic spine density, associated with an increase in spine length and a decrease in head width. These changes are associated with a decrease in F-actin levels within spine heads. On the other hand, a gain-of-function approach, applying exogenous BDNF, could not reproduce the increase in spine density or the changes in spine morphology previously described. Taken together, we show here that the effects exerted by BDNF on the dendritic architecture of hippocampal neurons are dependent on the neuron's maturation stage. Indeed, in mature hippocampal neurons in vitro as shown in vivo BDNF is specifically required for the activity-dependent maintenance of the mature spine phenotype.