Courses
Undergraduate Courses
Danny Eytan, together with Uri Shalit on machine learning in health care. It is an advanced undergrad/graduate course, mostly in seminar form. It was given in English in the past, depending on the student composition.
Machine Learning Has Great Potential For Improving Healthcare. We Will Read and Critically Discuss Recent Papers Covering How Ml Is Used For# Guiding Diagnosis, Personalizing Treatments, Analyzing Medical Images and Medical Text, Reducing Healthcare Costs, Reducing Unfair Bias, and For a Better Understanding of Human Physiology. Learning Outcomes# At The End of The Course The Student Will Know# 1. to Give Three Examples of Diagnostic Tasks in Which Machine Learning Tools Showed an Advantage Over Physicians 2. Explain How Machine Learning Tools Can Help Personalize Treatment, And What Are The Special Difficulties in This Task As Opposed To Diagnostic Tasks 3. Explain For Which Medical Tasks The Current Tools of Deep Learning Are Appropriate, and For Which Tasks They Are Not 4. Implement a Machine Learning Model On Medical Data, and Explain To A Medical Practitioner The Advantages and Disadvantages of The Model Versus a Human Physician,further, The Student Will Explain The Situations in Which They Recommend Using The Model, and The Situations in Which It Is Best to Avoid Using It 5. Raise a Number of Ethical Issues Arising From The Use of Machine Learning Tools in The Healthcare Domain
Faculty: Industrial Engineering and Management
|Undergraduate Studies |Graduate Studies
The Syllabus Will Be Announced at The Beginning of The Semeter.
Faculty: Physics
|Undergraduate Studies |Graduate Studies
Daniel Soudry is teaching the basic machine learning course. Given in Hebrew (joint undergrad).
An Introductory Course On Learning Systems in The Context Of Signal Processing, Artificial Intelligence and Control. Problems of Classification, Regression and Clustering. Neural Networks# Multi-level Perceptrons and Radial Basis Functions. Decision Trees. Elements of The Learning Theory# The Bayesian Approach, Hypothesis Spaces. Dimensionality Reduction Using Principal Components. Classification Using Support Vector Machines. Reinforcement Learning.
Faculty: Electrical and Computer Engineering
|Undergraduate Studies |Graduate Studies
מבוא לנושאי הפקת מידע מנתונים ושיטות למידה לא מפוקחת. יסודות בהסקה סטטיסטית: אמידה פרמטרית ולא-פרמטרית, בדיקת השערות. עיבוד ראשוני של נתונים. בחירת מאפיינים. שיטות להורדת מימדיות: ניתוח רכיבים עיקריים, פירוק ערכים סינגולריים, הרחבות לא-לינאריות. מדדי מרחק ודימיון בין פריטי מידע. אלגוריתמים לאשכול. זיהוי שכיחות וקשר, וזיהוי חריגים. יישומים מייצגים. תוצאות למידה: בתום הקורס הסטודנט יהיה מסוגל: 1. לתאר את בעיות היסוד בניתוח מידע. 2. להסביר וליישם שיטות סטטיסטיות להערכת פרמטרים ובדיקת השערות מתוך מידע 3. להסביר וליישם שיטות בסיסיות לבחירת מאפיינים. 4. להסביר וליישם אלגוריתמים להורדת ממדיות מידע. 5. להסביר וליישם אלגוריתמים לזיהוי שכיחות וקשר. 6. להסביר וליישם אלגוריתמים לאשכול נתונים. 7. להסביר וליישם אלגוריתמים לניתוח וזיהוי חריגים במידע.
Faculty: Electrical and Computer Engineering
|Undergraduate Studies |Graduate Studies
We Will Learn Theoretical and Practical Tools to Build, Design And Analyze Deep Networks, With an Emphasis On Supervised Learning. For Example, Properties and Covnergence of Gradient Desecnt and Its Variants, Efficient Differntiation, Multilayer Nets (approximation And Symmetry), Convnets (and Extentions) For Visual Tasks, Training Methods and Their Analysis, Networks For Serial Data, And Pre-training. Learning Outcomes# With The Completionof The Course, The Students# 1. Will Be Familiar With The Main Models and Common Training Methods For Deep Learning. 2. Will Be Able to Code (in Python, Using The Pytorch Framework) For Deep Neural Network, Train It, and Use It. 3. Will Be Able to Understand The Considerations Required to Tune Deep Networks For Achieving Good Perfomance, and The Relevant Theoretical Results (when Such Exist).
Faculty: Electrical and Computer Engineering
|Undergraduate Studies |Graduate Studies
Developing a Systems-level Mathematical Approach to The Single Neuron As a Prototype of Complex Biological Systems. Two Main Mathematical Persectives, Will Be Emphasized# Nonlinear Dyanmical Systems And Random Processes in Continuous Time. Main Topics# Introduction to The Biophysics of Neurons, Axons and Dendrites and Their Analyais As Input-output Systems in The Engineering Sense. The Neuron# Basic Properties of The Membrane and Its Chemical and Electrical Properties - Electrodiffusion and The Resting Potential. The Propagation of Signals Along Passive Cables and Cell to Cell Communication Through Chemical Synapses. The Hodgkin-huxley Model For Cell Excitbability, and Simplified Mathematical Models Allowing A Detailed Analysis of Excitability. Stochastic Elements in Neurons And Synapses.
Faculty: Electrical and Computer Engineering
|Undergraduate Studies |Graduate Studies
Graduate Courses
Ron Meir is giving the exploration/exploitation course, It is more geared towards Machine Learning, and requires some background in that field + probabilistic tools.
Autonomous Agents Operating in a Complex Environment Are Required To Function Under Challenging Conditions of Uncertainty, Resulting From Partial, Noisy and Delayed Information, From Lack of a World Model, And System Malfunction and Form Communication Bottlenecks. a Possible Approach to These Difficulties Combines Exploration and Exploitation. Roughly, Exploitation Involves Utilizing Prior Knowledge, Collected Through Activity Aimed at Achieving Required Gals, While Exploration Focuses On Searching For Modes of Operation With Potential Future Gains. The Optimal Balance Between Exploration and Exploitation Occupies a Basic Place in The The Areas of Optimal Control And Reinforcement Learning Since The Early 1960s, With Increasing Importance in Recent Years. in Spite of This, Except For Restricted Cases, There Is Limited Understanding of How to Balance The Two. In This Course We Will Characterize This Balance in Different Learning Systems, Aiming at The Extraction of General Principles, Opening The Door to The Development of Effective Exploration-exploitation Schemes in Challenging Problems in Machine Learning. Learning Outcomes# Understanding The Balance Between Exploration And Exploitation in Learning Stems, Understanding The Basic Theory For Simple Systems, Designing Effective an Exploration-exploitation Balance in More Complex Systems, Reading The Current Literature.
Faculty: Electrical and Computer Engineering
|Graduate Studies
Omri Barak, together with Jackie Schiller and Yoram Gutfreund is giving a neuro video-seminar course. Students watch an online lecture together, present a relevant paper and discuss it. Given in English.
Each Week The Students Will View a Video Lecture of a Prominent Neuroscientist. in The Seminar We Will Present and Discuss His Research. The Course Will Cover a Range Of Exciting Topics Such As Motor Control, Cortical Processing, Bird Song, Sensing, Consciousness and More. The Video Lectures Are Available Thanks to The Icnc, The Hebrew Univers
Faculty: Medicine
|Graduate Studies
Various Mathematical Models in Neuroscience. The Course Will Include Approximately Five Differents Models That Will Vary From Year to Year For Instance# Ring Model. Hopfield Model. Model of Grid Cells. Accumulation of Evidence. Context Dependent Integration of Evidence. Learning Outcomes# The End of The Course The Student Will Know# 1. Explain The Meaning of a Differential Equation 2. Numerically Simulate Ordinary Differential Equations. 3. Analyze a Dynamical System Using Basic Analytic Tools. 4. to Describe The Connection Between a Simple Mathematical Model and Experimental Results.