CS 732: Advanced machine learning
Spring 2019

Instructor: Usman Roshan
Office: GITC 4214B
Ph: 973-596-2872
Office hours: MW: 2 to 5
Email: usman@njit.edu

Textbook: Not required but following are recommended:

• Intro to Machine Learning by Ethem Alpaydin
• Learning with kernels by Scholkopf and Smola
• Foundations of Machine Learning by Mohri et. al.

Grading: 50% two paper presentations, 50% final project and presentation
Course Overview: We will focus on medical AI problems in this course. We will cover convolutional neural networks for MRI images, machine learning for ultrasound images, and adversarial attacks.

Course plan:

Topic	Date	Notes
Basic machine learning and Python scikit-learn	01/28/2019	Background Basic statistics More basic probability and statistics Applied statistics Linear algebra background More linear algebra Unix and login to NJIT machines Basic Unix command sheet Instructions for AFS login Basic machine learning background with Python scikit-learn Basic geometry of a linear classifier Empirical risk minimization Regularized risk minimization Regularization Support vector machine Datasets Linear data Non-linear data Breast cancer (bc.train.0) (bc.test.0) Ionosphere (ion.train.0) (ion.test.0) Climate simulation (climate.train.0) (climate.test.0) Qsar (qsar.train.0) (qsar.test.0) Fifty two datasets selected from the UCI machine learning repository Additional papers Representation learning Geometrical and Statistical properties of systems of linear inequalities with applications in pattern recognition (Cover 1965) Approximations by superpositions of sigmoidal functions (Cybenko 1989) Approximation Capabilities of Multilayer Feedforward Networks (Hornik 1991) ImageNet classification with deep neural networks (Krizhevsky et. al. 2012) Random projections preserve margin Random projections preserve margin II
Neural networks, image classification with machine learning, and convolutions	02/04/2019	Multi-layer perceptrons Scikit-learn MLP code Image classification code Python Image Library
Convolutional neural networks, training them with Keras	02/11/2019	Convolutional neural networks for image recognition Flower image classification with CNNs
Convolutions and paper presentations	02/18/2019	Breast cancer prediction with SVMs (Yajuan Li) Ultrasound image classification (Cheng Zhong) Convolutional kernels
Paper presentations	02/25/2019	Dual-force convolutional neural networks for accurate brain tumor segmentation (Hadi) DeepPolyA: A Convolutional Neural Network Approach for Polyadenylation Site Prediction (John Chen)
Paper presentations	03/04/2019	DeepPolyA: A Convolutional Neural Network Approach for Polyadenylation Site Prediction (John Chen) An overview of deep learning in medical imaging focusing on MRI (Yanan Yang) Visualizing Deep Neural Network Decisions:Prediction Difference Analysis (Michael Lan)
Paper presentations	03/18/2019 (rescheduled from 03/11/19)	Motion Corrected Multishot MRI Reconstruction Using Generative Networks with Sensitivity Encoding (Craig Kenney) Convolutional neural network based Alzheimer’s disease classification from magnetic resonance brain images (Ruoyu) Heartbeat anomaly detection using adversarial oversampling (Syed) Prediction of lung cancer patient survival via supervised machine learning classification techniques (Firas)
Paper presentations	03/25/2019	Joint CS-MRI Reconstruction and Segmentation with a Unified Deep Network (Xiangyu) Predicting Alzheimer's disease: a neuroimaging study with 3D convolutional neural networks (Kannan) Digital mammographic tumor classification using transfer learning from deep convolutional neural networks3D (Jeremy)
Paper presentations	04/01/2019	Decision tree and random forest models for outcome prediction in antibody incompatible kidney transplantation (Xiaopeng) A CNN for the Automatic Diagnosis of Collagen VI related Muscular Dystrophies (Jiali) LightGBM: A Highly Efficient Gradient Boosting Decision Tree (Xiaowei) Image classification v2 CIFAR 10 CIFAR 100 STL 10 Mini ImageNet
Paper presentations	04/08/2019	Random Forest ensembles for detection and prediction of Alzheimer's disease with a good between-cohort robustness (Hadi) Automatic segmentation method of pelvic floor levator hiatus in ultrasound using a self-normalising neural network (Cheng) Risk prediction for breast Cancer in Han Chinese women based on a cause-specific Hazrad model (Jiankai) Joint Learning of Words and Meaning Representation for Open-Text Semantic Parsing (Chih-yuan)
Paper presentations	04/15/2019	Asymmetric Loss Functions and Deep Densely- Connected Networks for Highly-Imbalanced Medical Image Segmentation: Application to Multiple Sclerosis Lesion Detection (Dan) Deep learning models for bateria taxonomic classification of metagenomic data (John) Generalizing multistain immunohistochemistry tissue segmentation using one-shot color deconvolution deep neural networks (Michael) CheXNet: Radiologist-Level Pneumonia Detection on Chest X-Rays with Deep Learning (Yanan) NiftyNet: A Deep learning platform for medical Imaging (Syed)
Paper presentations, transposed convolutions	04/22/2019	Learning Deep CNN Denoiser Prior for Image Restoration (Yajuan) Efficient Parameter-free Clustering Using First Neighbor Relations (Craig) Automatic segmentation of the spinal cord and intramedullary multiple sclerosis lesions with convolutional neural networks (Ruoyo) Melanoma detection by analysis of clinical images using convolutional neural network (Firas) Firas projects results (Firas) Deep Convolutional Neural Networks for Breast Cancer Histology Image Analysis (Honghao) A guide to convolution arithmetic for deep learning
Paper presentations, transposed convolutions	05/04/2019	Project results: Cheng, Firas, Yanan Paper presentation: Xiaowei, Jiankai, Jiali, Xiangyu, Chiy-yuan, Dan, Jeremy A Fully-Automatic Framework for Parkinson’s Disease Diagnosis by Multi-Modality Images (Xiangyu) AnomiGAN: Generative adversarial networks for anonymizing private medical data (Jiali) Distributed Representation of Words and Phrases and their Compositionality (Chih-yuan) Optimus: An Efficient Dynamic Resource Scheduler for Deep Learning Clusters (Xiaowei) Yanan's project Cheng's project Syed's project
Paper and project presentations	05/06/2019	Automated Classification of Lung Cancer Types from Cytological Images Using Deep Convolutional Neural Networks (Jeremy) Colorectal\ cancer\ diagnosis\ from\ histology\ images\ A\ comparative\ study (Honghao) 3D Deep Learning for Efficient and Robust Landmark Detection in Volumetric Data (Kannan) Classification of Dynamic Contrast Enhanced MR Images of Cervical Cancers Using Texture Analysis and Support Vector Machines (Dan) Deep learning based classification of focal liver lesions with contrast-enhanced (Xiaopeng) Forecasting Lung Cancer Diagnoses with Deep Learning (Jiankai) Projects: Michael Craig Hadi Firas
Projects and papers		Hadi: (a) Dual-force convolutional neural networks for accurate brain tumor segmentation (b) Random Forest ensembles for detection and prediction of Alzheimer3s disease with a good between-cohort robustness Michael Lan: (a) Visualizing Deep Neural Network Decisions: Prediction Difference Analysis (b) Cheng Zhong: (a) An improved deep learning approach for detection of thyroid papillary cancer in ultrasound images (b) Automatic segmentation method of pelvic floor levator hiatus in ultrasound using a self-normalising neural network Yanan Yang: (a) An overview of deep learning in medical imaging focusing on MRI (b) CheXNet: Radiologist-Level Pneumonia Detection on Chest X-Rays with Deep Learning John Chen: (a) DeepPolyA: A Convolutional Neural Network Approach for Polyadenylation Site Prediction (b) DeeReCT-PolyA: a robust and generic deep learning method for PAS identification Yajuan Li: (a) SVM and SVM Ensembles in Breast Cancer Prediction (b) Learning Deep CNN Denoiser Prior for Image Restoration Xiaopeng Jiang: Decision tree and random forest models for outcome prediction in antibody incompatible kidney transplantation (b) Firas Gerges: (a) Prediction of lung cancer patient survival via supervised machine learning classification techniques (b) Melanoma detection by analysis of clinical images using convolutional neural network Ruoyo Wang: (a) Convolutional neural network based Alzheimer’s disease classification from magnetic resonance brain images (b) Automatic segmentation of the spinal cord and intramedullary multiple sclerosis lesions with convolutional neural networks Author links open overlay panel Jiankai Li: (a) Risk prediction for breast Cancer in Han Chinese women based on a cause-specific Hazard model (b) Jiali Wang: (a) A Convolutional Neural Network for the Automatic Diagnosis of Collagen VI related Muscular Dystrophies (b) AnomiGAN: Generative adversarial networks for anonymizing private medical data Craig: (a) Motion Corrected Multishot MRI Reconstruction Using Generative Networks with Sensitivity Encoding (b) Efficient Parameter-free Clustering Using First Neighbor Relations Syed: (a) Heartbeat Anomaly Detection using Adversarial Oversampling (b) Character-level Convolutional Networks for Text Classification Xiangyu Gao: (a) Joint CS-MRI Reconstruction and Segmentation with a Unified Deep Network (b) A Fully-Automatic Framework for Parkinson's Disease Diagnosis by Multi-Modality Images Kannan: (a) 3D Deep Learning for Efficient and Robust Landmark Detection in Volumetric Data (b) Predicting Alzheimer's disease: a neuroimaging study with 3D convolutional neural networks Jeremy Hui: (a) Digital mammographic tumor classification using transfer learning from deep convolutional neural networks (b) Automated Classification of Lung Cancer Types from Cytological Images Using Deep Convolutional Neural Networks Chih-yuan (a) Joint Learning of Words and Meaning Representations for Open-Text Semantic Parsing (b) Dan (a) Asymmetric Loss Functions and Deep Densely- Connected Networks for Highly-Imbalanced Medical Image Segmentation: Application to Multiple Sclerosis Lesion Detection (b) Dan (a) Asymmetric Loss Functions and Deep Densely- Connected Networks for Highly-Imbalanced Medical Image Segmentation: Application to Multiple Sclerosis Lesion Detection (b) Honghao (a) Deep Convolutional Neural Networks for Breast Cancer Histology Image Analysis (b) Open papers: Heartbeat Anomaly Detection using Adversarial OversamplingM Strong Black-box Adversarial Attacks on Unsupervised Machine Learning Models SCATGAN FOR RECONSTRUCTION OF ULTRASOUND SCATTERERS USING GENERATIVE ADVERSARIAL NETWORKS Similarity-Based Retrieval for Biomedical Applications Content-Based and Similarity-Based Querying for Broad-Usage Medical Image Retrieval Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks
Medical AI		Brain MRI machine learning papers: Machine learning for brain MRIs Convolutional neural networks for brain MRIs UNets for masking images Public ATLAS dataset (see ATLAS) Other MRI machine learning studies Cancer prediction papers: Cancer case and control prediction Cancer survival time prediction Cancer prediction from image data (see TCGA portal) Ultrasound image machine learning papers: Ultrasound image classification Disease risk prediction: Heart disease, type I and II diabetes, and other general disease risk prediction