University Guest Lectures

Resiliency and Operations Lessons from the Apollo Missions to the Moon

July 2019 marked 50 years since the first humans set foot on the moon.

Moments before the historic landing of Apollo 11, the spacecraft computer started throwing errors and basically restarting every few seconds. The astronauts and mission control had to …

Securing the Crown Jewels

Checking every event will impact the mission-critical-tasks and the company's revenue, on the other hand missing one suspicious event may cause catastrophe to the organization and the market.
I will talk about technique to "find the needle in the haystack", …

Self-Supervised Learning from Videos

A vast amount of audio-visual data is available on the Internet thanks to video streaming services, to which users upload their content. However, there are difficulties in exploiting this data for supervised learning due to the
lack of labels. More specifically, …

Modernizing ITSM - Cloud Service Management and Operations (CSMO)

With the evolving landscape towards DevOps and cloud, IT Service Management (ITSM) has to change as well to stay relevant.

Cloud Service Management and Operations (CSMO) refers to the entirety of activities – organized in persona, processes, and tools – …

Intellectual Property for Engineers and Start-ups

Intellectual property (IP) has value, whether it be the copyright, trademark, invention or confidential information. As a person, and as a company, it is important to understand the ins and outs of IP. In this lecture we will introduce the …

Constraint Satisfaction Problems: From theory to products

Constraint Satisfaction Problems (CSP) is a field of artificial intelligence (AI) that helps solve real-life problems by specifying (or modeling) the legal solution structure, rather than implementing algorithms that solve these problems. IBM’s CSP solving system declares the solution model …

Big Data Analytics with Spark

I will explain the challenges of current Big Data engines ( in particular Apache Spark ) integration with object stores, what are the issues and their origin. I will discuss what can be done to make this integration more efficient …

Compression and deduplication in Storage Systems

With the growing amount of data being used and stored, storage is challenged to keep up with the capacity demands. Modern storage systems have hence adopted various data reduction techniques, mainly the use of loss-less compression and deduplication techniques. In …

Cognitive Marketing

Artificial intelligence (AI) started out as a field of study at MIT in the '50s, focused on getting computers to act like humans. Today, AI is helping humans get a jump on everything from data analysis to decision-making and insight …

Agile, don't just do. Be.

The methodology of Agile, originally used in Software Development, has spread its wings, as business community starts to talk about Business Agility and Agile leaders.
The purpose of this lecture is to provide an initial introduction to the world of Agile, …

Agile – don't just do. Be.

The methodology of Agile, originally used in Software Development, has spread its wings, as business community starts to talk about Business Agility and Agile leaders.
The purpose of this lecture is to provide an initial introduction to the world of Agile, …

Speech synthesis with voice transformations

A text-to-speech synthesis (TTS) system can speak in few voices, each is derived from audio recordings of a real person. TTS voice transformations that change a perceived speaker identity in a controllable way is an attractive alternative to expensive, lengthy …

Blockchain is ready for Business

IBM is interested in exploring the broader business application of Blockchain technology. This is a transformational opportunity for many of IBM's clients since it touches most of the industries. In this presentation we will explain IBM Blockchain offering, deep dive …

Are you ready for the Digital Era?

Do the Right things, Do the things Right, Do it Right the first time!
Digital is the new Holy Grail. The battleground between companies has moved from the physical world to the Digital one and Transformation is the Enabler for winning …

IoT Employee Safety solution

How a First-of-a-kind initiative that was conceived in Haifa has led us to develop and implement the Employee Safety solution prototype that was deployed at a client site in Canada where temperature drops down to below-freezing level of -40C.
In this …

QRadar Apps

The proposed talk provides a practical introduction to the development of interactive security applications, using a specialized framework designed by IBM to help overcome the ever-growing number of cyber-attacks on commercial and government computer infrastructures.

The business problem – securing corporate …

How Do You Like Your Phish?

Phishing is one of the early, simple and cost-effective scams in the internet.
Using Social engineering, an attacker aims to acquire sensitive information like passwords and credit card details, for malicious reasons. The attacker is doing so by masquerading as a …

Securing (?) Internet Connections

The Internet – with 2.5 billion users – was never designed to be a secure system. Protocols like DNS or HTTP, the main building blocks of the Internet, were not designed with security in mind. Security protocols and extensions, such …

Trust services with Hardware support (Intel SGX)

The talk will describe new paradigms for securely running code on commodity hardware, with a focus on the Intel SGX technology. I will describe the basic concepts, the main use-cases and deployment challenges.

A deep dive into Biometrics

With the rapid growth in the use of mobile devices and pervasive computing on one hand, and the rise of deep learning on the other, biometric recognition has become an even more exciting field. In this talk, I will give …

Data Center Network Virtualization for the Cloud

This is an introduction to cloud connectivity challenges and solutions. First, the lecture describes why and how networking in cloud environments differs from the traditional networking. Next, it presents specific requirements for connectivity in the cloud and gives a generic …

What's new in IBM Cloud Functions Land?

Serverless computing or FaaS (Function as a Service) provides developers with a "server less" deployment and operational model relieving them from the need to worry about complex infrastructure and operational tasks. FaaS enables developers to focus on developing and executing …

Cloud object stores: distributed storage for the cloud

I will describe Object Stores, which replaced the traditional (POSIX) file systems to store large volumes of new data types created in the Cloud era, e.g. text, video/image, CSV logs. Object stores are distributed, very scalable and cost-effective storage systems, …

Efficient SQL Analytics on Big Data in Object Storage

Big datasets such as IoT device and server logs are increasingly stored in object storage to achieve low cost and endless capacity. This talk will discuss techniques for efficient SQL analytics of these datasets directly on object storage using Apache …

Function as a service: a new cloud programming model

Function as a service (FaaS) is the latest major addition to the already complex and diverse cloud computing landscape. Initially introduced by Amazon in 2014, it now rapidly evolves by vendors and open source projects. I will describe FaaS, its …

From Neural Networks to Deep Learning

Artificial intelligence has become the hottest talk of the 21st century, bringing us self- driving cars, speech recognition and early disease diagnosis. Deep Learning specifically, has become one of the most popular buzz-words in the field. Deep Learning capabilities are …

Large Scale Labelling Methodology

Machine learning is the science of getting computers to act without being explicitly programmed. While traditional software solutions rely on sets of rules for executing an algorithm in order to solve a problem, machine learning based systems rely on given …

Creating a smart diagnosis system

There are many cases in which experts are required to carry out diagnosis, i.e., given a list of observed symptoms, find the problem that is causing these symptoms. Examples include medical diagnosis, where a healthcare professional needs to discern the …

Creating Smart Personal Assistants

There is a proliferation of products being labeled as smart personal assistants, including Siri, Amazon Echo and Google Assistant. However, currently, such assistants are quite limited, and are able to carry out interact with users on a small number of …

Detecting and expressing affect in text

Humans experience of the world can be decomposed into Cognition (what we think), Affects (what we feel) and Action (what we do). Until now, computers have been very successful in the Cognitive and Action domains: computers replace, extend and analyze …

Big Data Analytics with Apache Spark

Apache Spark is an open-source scalable in-memory computation framework. It is widely used across the industry and supports a variety Big Data Analytics use cases. In this talk we describe Spark's main components, its rich set of libraries and operation …

Unsupervised Summarization Techniques

The abundance of unstructured information raises the need for automatic systems that can “condense” information from various documents into a shorter length, readable summary. Such summaries may further be required to cover a specific information need (e.g., summarizing web search …

Video Scene Detection

Video scene detection is the task of dividing a video into semantic sections. I will present our novel and effective method for temporal grouping of scenes using an arbitrary set of features computed from the video. This task is formulated …

A Deep Learning emotion predictor from speech signals

Statistical (machine learning) methods have been applied successfully to speech signals in order to predict the emotional state from the nonverbal speech content. Common approach is based on extracting features that were shown to represent the required information (emotions), and …

Text dependent speaker recognition

In this talk I will give an overview on text dependent speaker recognition including Bayesian and deep-learning based approaches. In the Bayesian part I will describe methods tuned to handling large amounts of training data (joint-factor analysis, i- vectors) and …

Inter-Dataset Variability Modeling for Speaker Recognition

Speaker recognition in a mismatched domain was the main focus of the recent NIST speaker recognition evaluation. In my talk I will introduce a novel approach for addressing this challenge. The main principle is to learn the inter-dataset variability in …

Challenges in the Chatbot

Chatbots will probably predominantly own the future of man-machine interface. We see them progressively taking over in Customer Care centers, Command and Control systems, and as Personal assistance. From the underlying framework point of view, one of the most interesting …

Neural Networks with External-Memory

In this lecture we will discuss architectures which extend the capabilities of neural networks by coupling them to external memory resources, which they can interact with using the attention mechanism. We will review models from the domains of language understanding …