yiyg510 Goto Github PK

a_v_uncertain

Code for our ISBI 2019 paper on Artery/Vein classification with uncertainty predictions

attention-driven-loss

Attention-Driven Loss for Anomaly Detection in Video Surveillance

autoretinalimagesegmentation

This code is used for joint optic disc and cup segmentation from retinal fundus images

av-segmentation

awc

CFEA: Collaborative Feature Ensembling Adaptation for Domain Adaptation in Unsupervised Optic Disc and Cup Segmentation

awesome-ml-for-cybersecurity

:octocat: Machine Learning for Cyber Security

awesome-semi-supervised-learning

:scroll: An up-to-date & curated list of awesome semi-supervised learning papers, methods & resources.

basnet

Code for CVPR 2019 paper. BASNet: Boundary-Aware Salient Object Detection

bcdu-net

BCDU-Net : Medical Image Segmentation

beal

code for paper Boundary and Entropy-driven Adversarial Learning for Fundus Image Segmentation

bi-lstm-with-attention-mechanism

Woodblock-printing Mongolian Words Recognition Method

bibm2020-hynet-for-optic-disc-segmentation

Codes for "An Efficient Weakly-Supervised Learning Method for Optic Disc Segmentation" published in IEEE BIBM2020

bioimage-classification-with-handcrafted-and-learned-features-

blood-vessels-segmentation

Retinal blood vessel segmentation using 3 approaches: image processing techniques, machine learning with LightGBM and deeplearning with convolutional neural network (U-net)

book_deeplearning_in_pytorch_source

brain-segmentation

brain-tumor-anomaly-detection-and-segmentation

TEAM: Aasha, Ronak, Chinmay, Manmeet

bseresu-net

Tensorflow implementation of BSEResU-Net: An attention-based before-activation residual U-Net for retinal vessel segmentation

c2ae-multilabel-classification

Tensorflow implementation for the paper 'Learning Deep Latent Spaces for Multi-Label Classfications' in AAAI 2017

captcha_generator

CAPTCHA generator using cv2. Just for fun :)

cat-recognition-app

The repository include code for build a ML application using Flask, Tensorflow

cat-recognition-train

Train and deploy a cat vs dog image recognition model using TensorFlow

cca_fusion

Code accompanying our ISBI 2021 paper "Multimodal Fusion using Sparse CCA for Breast Cancer Survival Prediction" https://arxiv.org/abs/2103.05432 by Vaishnavi Subramanian, Tanveer Syeda-Mahmood, Minh N. Do

ce-net

The manuscript has been accepted in TMI.

cfl_for_mmif

Matlab codes - Multimodal Image Fusion via Coupled Feature Learning

cfs

Curvature-based Feature Selection method

chandlerzuo.github.io

My blog.

classification_for_human_resources

Using Deep Learning for a Human Resources Department in order to predict which employees are more likely to quit.

classify-websites-into-amiable-or-malicious-using-raw-data

With WWW being the global platform, various fields inclusive to the same have emerged until hitherto. Due to ever-changing forms of cyber Security, it has become a necessity to classify Malicious websites so as to secure personal content. In this project we have implemented The State-Of-the-Art Decision Tree Machine Learning Models such as Random Forest and Decision Tree to classify URLs as malicious or amiable. Implementation of Classification algorithms for discrete data as well as normal regression model is used in the project. Malevolent URLs have been broadly used to mount different digital assaults including spamming, phishing and malware. Recognition of malignant URLs and distinguishing proof of danger types are basic to upset these assaults. Knowing the sort of a danger empowers assessment of seriousness of the assault and embraces a viable countermeasure. Existing strategies commonly distinguish vindictive URLs of a solitary assault type. In this paper, we propose technique utilizing AI to identify malevolent URLs of all the mainstream assault types. While the World Wide Web has become a stellar application on the Internet, it has likewise gotten a massive danger of digital assaults. Enemies have utilized the Web as a vehicle to convey malignant assaults, for example, phishing, spamming, and malware contamination. For instance, phishing ordinarily includes sending an email apparently from a dependable source to deceive individuals to click a URL (Uniform Resource Locator) contained in the email that joins to a fake page. To address Web-based assaults, an incredible exertion has been coordinated towards identification of noxious URLs. A typical countermeasure is to utilize a boycott of vindictive URLs, which can be built from different sources, particularly human criticisms that are exceptionally precise yet tedious. Boycotting acquires no bogus positives, however is successful just for known noxious URLs. It can't identify obscure malevolent URLs. The very idea of careful match in boycotting these renders it simple to be sidestepped. This shortcoming of blacklisting has been tended to by oddity-based location techniques intended to identify obscure vindictive URLs. In these strategies, a characterization model dependent on discriminative principles or highlights is worked with either information from the earlier or through machine learning. Choice of discriminative standards or highlights assumes a basic function for the presentation of a locator. Online malware assaults become one in everything about chief genuine dangers that need to be tended too frantically. Numerous methodologies that have stood out as promising manners by which of safeguard work, for example, malware grasp utilizing various boycotts. Nonetheless, these standard methodologies ordinarily neglect to watch new assaults due to the adaptability of malignant sites. Consequently, it's hard to deal with state-of-the-art boycotts with data concerning new vindictive sites. Malignant location identification assumes a significant part for a few network protection applications, and unmistakably AI moves toward square measure a promising course. In mix with protection imperatives on information sets of real client traffic, its irksome for scientists and product engineers to measure hostile to malware arrangements against huge scope information sets of practical net traffic. AI strategy [1] region unit utilized so as to characterize the online deals into malignant and benevolent URLs. The appearance of ongoing correspondence innovations has had enormous contact with in the development and advancement of organizations spamming over a few applications just as web based banking, online business, and long range informal communication. In actuality, in the present age it's almost required to have a web presence to run a famous endeavour. Accordingly, the significance of the overall net has ceaselessly been expanding. Unfortunately the mechanical promotions return in expansion to new unobtrusive strategies to assault and trick client. Such assaults grasp noxious sites that sell fake stock, financial extortion by fooling clients into uncovering delicate data that in the long run cause stealing of money or character, or maybe placing in malware inside the clients framework. There square measure a huge kind of procedures to actualize such assaults, similar to explicit hacking attempts, Derive-by abuses, Denial of administration [2], Distributed refusal of administration [1] and bunches of others. Concentrating the changeability of assaults, without a doubt new assault assortments, and furthermore the unnumbered settings inside which such assaults will appears, it's exhausting to style-solid frameworks to find digital security penetrates. The restrictions of customary security the board advancements are getting to an ever increasing extent genuine given this remarkable development of new security dangers, fast changes of new IT advancements, and critical deficiency of security experts. The vast majority of these assaulting strategies are acknowledged through spreading traded off URLs. A primary exploration exertion in pernicious URL recognition has zeroed in on choosing profoundly successful discriminative highlights. Existing techniques were intended to distinguish pernicious URLs of a solitary assault type, for example, spamming, phishing, or malware. In this paper, we propose a strategy utilizing Machine Learning Algorithms on how to distinguish malevolent URLs of all the well known assault types including phishing, spamming and malware contamination, and distinguish the assault types noxious URLs endeavour to dispatch.

cnafs

Convex Nonnegative Matrix Factorization with Adaptive Graph for Unsupervised Feature Selection