Name: The Ways Machine Learning and AI Can Fail
Uploaded: 2017-04-13T17:47:56.000Z
Duration: 47 min 56 s
Description: Good applications of machine learning and AI can be difficult to pull off. Join Brian Lange, Partner and Data Scientist at data science firm Datascope, as he discusses a variety of ways machine learning and AI can fail (from technical to human factors) so that you can avoid repeating them yourself.

Implementing AI applications based on machine learning is a significant topic for organizations embracing digital transformation. By 2020, 30% of CIOs will include AI in their top five investment priorities according to Gartner’s Top 10 Strategic Technology Trends for 2018: Intelligent Apps and Analytics. But to deliver on the AI promise, organizations need to generate good quality data to train the algorithms. Failure to do so will result in the following scenario: "When you automate a mess, you get an automated mess."

This webinar covers: 

- An introduction to machine learning use cases and challenges provided by Kirk Borne, Principal Data Scientist at Booz Allen Hamilton and top data science and big data influencer.
- How to achieve good data quality based on harmonized semantic metadata presented by Andreas Blumauer, CEO and co-founder of Semantic Web Company and a pioneer in the application of semantic web standards for enterprise data integration.
- How to apply a combined approach when semantic knowledge models and machine learning build the basis of your cognitive computing. (See Attachment: The Knowledge Graph as the Default Data Model for Machine Learning)
- Why a combination of machine and human computation approaches is required, not only from an ethical but also from a technical perspective.

Semantic AI: Bringing Machine Learning and Knowledge Graphs Together

The 10x growth of transaction volumes, 50x growth in data volumes and drive for real-time visibility and responsiveness over the last decade have pushed traditional technologies including databases beyond their limits. Your choices are either buy expensive hardware to accelerate the wrong architecture, or do what other companies have started to do and invest in technologies being used for modern hybrid transactional analytical applications (HTAP). 

Learn some of the current best practices in building HTAP applications, and the differences between two of the more common technologies companies use: Apache® Cassandra™ and Apache® Ignite™. This session will cover:

- The requirements for real-time, high volume HTAP applications 
- Architectural best practices, including how in-memory computing fits in and has eliminated tradeoffs between consistency, speed and scale
- A detailed comparison of Apache Ignite and GridGain® for HTAP applications

About the speaker: Denis Magda is the Director of Product Management at GridGain Systems, and Vice President of the Apache Ignite PMC. He is an expert in distributed systems and platforms who actively contributes to Apache Ignite and helps companies and individuals deploy it for mission-critical applications. You can be sure to come across Denis at conferences, workshop and other events sharing his knowledge about use case, best practices, and implementation tips and tricks on how to build efficient applications with in-memory data grids, distributed databases and in-memory computing platforms including Apache Ignite and GridGain.  

Before joining GridGain and becoming a part of Apache Ignite community, Denis worked for Oracle where he led the Java ME Embedded Porting Team -- helping bring Java to IoT.

Comparing Apache Ignite & Cassandra for Hybrid Transactional Analytical Apps

When monitoring an increasing number of machines, the infrastructure and tools need to be rethinked. A new tool, ExDeMon, for detecting anomalies and raising actions, has been developed to perform well on this growing infrastructure. Considerations of the development and implementation will be shared.

Daniel has been working at CERN for more than 3 years as Big Data developer, he has been implementing different tools for monitoring the computing infrastructure in the organisation.

Scalable Monitoring for the Growing CERN Infrastructure

As data analytics becomes more embedded within organizations, as an enterprise business practice, the methods and principles of agile processes must also be employed.

Agile includes DataOps, which refers to the tight coupling of data science model-building and model deployment. Agile can also refer to the rapid integration of new data sets into your big data environment for "zero-day" discovery, insights, and actionable intelligence.

The Data Lake is an advantageous approach to implementing an agile data environment, primarily because of its focus on "schema-on-read", thereby skipping the laborious, time-consuming, and fragile process of database modeling, refactoring, and re-indexing every time a new data set is ingested.

Another huge advantage of the data lake approach is the ability to annotate data sets and data granules with intelligent, searchable, reusable, flexible, user-generated, semantic, and contextual metatags. This tag layer makes your data "smart" -- and that makes your agile big data environment smart also!

The Data Lake for Agile Ingest, Discovery, & Analytics in Big Data Environments

With new technologies such as Hive LLAP or Spark SQL, do you still need a data warehouse or can you just put everything in a data lake and report off of that? No! In the presentation, James will discuss why you still need a relational data warehouse and how to use a data lake and an RDBMS data warehouse to get the best of both worlds.

James will go into detail on the characteristics of a data lake and its benefits and why you still need data governance tasks in a data lake. He'll also discuss using Hadoop as the data lake, data virtualization, and the need for OLAP in a big data solution, and he will put it all together by showing common big data architectures.

Is the Traditional Data Warehouse Dead?

HDFS on Kubernetes: Lessons Learned is a webinar presentation intended for software engineers, developers, and technical leads who develop Spark applications and are interested in running Spark on Kubernetes. Pepperdata has been exploring Kubernetes as potential Big Data platform with several other companies as part of a joint open source project.

In this webinar, Kimoon Kim will show you how to: 

–Run Spark application natively on Kubernetes
–Enable Spark on Kubernetes read and write data securely on HDFS protected by Kerberos

HDFS on Kubernetes: Lessons Learned

There has been a flood of publicity around big data, data processing, and the role of predictive analytics in businesses of the future.
As business operators how do we get access to these valuable business insights, even when there is not a data analyst around to walk us through their results? 

- Should your software emulate a data scientist?
- Learn about the power of data visualizations. 
- Learn about creating value from disperse data sets.

Data scientists: Can't live with them, can't live without them.

Fraud detection is a classic adversarial analytics challenge: As soon as an automated system successfully learns to stop one scheme, fraudsters move on to attack another way. Each scheme requires looking for different signals (i.e. features) to catch; is relatively rare (one in millions for finance or e-commerce); and may take months to investigate a single case (in healthcare or tax, for example) – making quality training data scarce.

This talk will cover a code walk-through, the key lessons learned while building such real-world software systems over the past few years. We'll look for fraud signals in public email datasets, using IPython and popular open-source libraries (scikit-learn, statsmodel, nltk, etc.) for data science and Apache Spark as the compute engine for scalable parallel processing.

David will iteratively build a machine-learned hybrid model – combining features from different data sources and algorithmic approaches, to catch diverse aspects of suspect behavior:

- Natural language processing: finding keywords in relevant context within unstructured text
- Statistical NLP: sentiment analysis via supervised machine learning
- Time series analysis: understanding daily/weekly cycles and changes in habitual behavior
- Graph analysis: finding actions outside the usual or expected network of people
- Heuristic rules: finding suspect actions based on past schemes or external datasets
- Topic modeling: highlighting use of keywords outside an expected context
- Anomaly detection: Fully unsupervised ranking of unusual behavior

Apache Spark is used to run these models at scale – in batch mode for model training and with Spark Streaming for production use. We’ll discuss the data model, computation, and feedback workflows, as well as some tools and libraries built on top of the open-source components to enable faster experimentation, optimization, and productization of the models.

Hunting Criminals with Hybrid Analytics, Semi-supervised Learning, & Feedback

In this talk, Ram will provide a unified framework for Internet of Things, Cyber-Physical Systems, and Smart Networked Systems and Societies, and then discuss the role of ontologies for interoperability.

The Internet, which has spanned several networks in a wide variety of domains, is having a significant impact on every aspect of our lives. These networks are currently being extended to have significant sensing capabilities, with the evolution of the Internet of Things (IoT). With additional control, we are entering the era of Cyber-physical Systems (CPS). In the near future, the networks will go beyond physically linked computers to include multimodal-information from biological, cognitive, semantic, and social networks.

This paradigm shift will involve symbiotic networks of people (social networks), smart devices, and smartphones or mobile personal computing and communication devices that will form smart net-centric systems and societies (SNSS) or Internet of Everything. These devices – and the network -- will be constantly sensing, monitoring, interpreting, and controlling the environment. 

A key technical challenge for realizing SNSS/IoE is that the network consists of things (both devices & humans) which are heterogeneous, yet need to be interoperable. In other words, devices and people need to interoperate in a seamless manner. This requires the development of standard terminologies (or ontologies) which capture the meaning and relations of objects and events. Creating and testing such terminologies will aid in effective recognition and reaction in a network-centric situation awareness environment.

Before joining the Software and Systems Division (his current position), Ram was the leader of the Design and Process group in the Manufacturing Systems Integration Division, Manufacturing Engineering Lab, where he conducted research on standards for interoperability of computer-aided design systems.

Toward Internet of Everything: Architectures, Standards, & Interoperability

If a volcano erupts in Iceland, why is Hong Kong your first supply chain casualty? And how do you figure out the most efficient route for bike share replacements? 

In this presentation, Chief Data Scientist Dmitri Adler will walk you through some of the most successful use cases of supply-chain management, the best practices for evaluating your supply chain, and how you can implement these strategies in your business.

Logistics Analytics: Predicting Supply-Chain Disruptions

Continuous streams of data are generated in every industry from sensors, IoT devices, business transactions, social media, network devices, clickstream logs etc. Within these streams of data lie insights that are waiting to be unlocked. 

This session with several live demonstrations will detail the build out of an end-to-end solution for the Internet of Things to transform data into insight, prediction, and action using cloud services. These cloud services enable you to quickly and easily build solutions to unlock insights, predict future trends, and take actions in near real-time.

Samartha (Sam) Chandrashekar is a Program Manager at Microsoft. He works on cloud services to enable machine learning and advanced analytics on streaming data.

Unlock real-time predictive insights from the Internet of Things

If a database is filled automatically, but it's not analyzed, can it make an impact? And how do you combine disparate data sources to give you a real-time look at your environment? 

Chief Executive Officer Merav Yuravlivker discusses how companies are missing out on some of their biggest profits (and how some companies are making billions) by aggregating disparate data sources. You'll learn about data sources available to you, how you can start automating this data collection, and the many insights that are at your fingertips.

Bridging the Data Silos

Data scientists know, the visualization of data doesn't materialize out of thin air, unfortunately. One of the most vital preparation tactics and dangerous moments happens in the ETL process.

Join Ray to learn the best strategies that lead to successful ETL and data visualization. He'll cover the following and what it means for visualization:

1. Data at Different Levels of Detail
2. Dirty Data
3. Restartability
4. Processing Considerations
5. Incremental Loading

Ray Rashid is a Senior Business Intelligence Consultant at Unilytics, specializing in ETL, data warehousing, data optimization, and data visualization. He has expertise in the financial, manufacturing and pharmaceutical industries.

Strategies for Successful Data Preparation

Agenda: 
Data warehousing at a glance
Wild West vs Enterprise HPE ALM Template
Planning and configuring the template
Customer use case: Standardization project walkthrough
How to maintain a standardized environment
Next steps with HPE ALM

HPE ALM Standardization as a Precursor for Data Warehousing

We all are aware of the challenges enterprises are having with growing data and silo’d data stores. Businesses are not able to make reliable decisions with un-trusted data and on top of that, they don’t have access to all data within and outside their enterprise to stay ahead of the competition and make key decisions for their business.

This session will take a deep dive into current Healthcare challenges businesses are having today, as well as, how to build a Modern Data Architecture using emerging technologies such as Hadoop, Spark, NoSQL datastores, MPP Data stores and scalable and cost effective cloud solutions such as AWS, Azure and BigStep.

Building Enterprise Scale Solutions for Healthcare with Modern Data Architecture

Past infrastructures provided compute, storage and network enabling static enterprise deployments which changed every few years. This talk will analyze the consequences of a world where production SAP and Spark clusters including data can be provisioned in minutes with the push of a button. 

What does it mean for the IT architecture of an enterprise? How to stay in control in a super agile world?

Data at the corner of SAP and AWS

Whether you're just starting out or a seasoned solution architect, developer, or data scientist, there are most likely key mistakes that you've probably made in the past, may be making now, or will most likely make in the future. In fact, these same mistakes are most likely impacting your company's overall success with their analytics program.

Join us for our upcoming webinar, 3 Critical Data Preparation Mistakes and How to avoid them, as we discuss 3 of the most critical, fundamental pitfalls and more!

• Importance of early and effective business partner engagement
• Importance of business context to governance
• Importance of change and learning to your development methodology

3 Critical Data Preparation Mistakes and How-to Avoid them

Good applications of machine learning and AI can be difficult to pull off. Join Brian Lange, Partner and Data Scientist at data science firm Datascope, as he discusses a variety of ways machine learning and AI can fail (from technical to human factors) so that you can avoid repeating them yourself.

The Ways Machine Learning and AI Can Fail

machine learning

data

data analytics

The data center management community focuses on the holistic management and optimization of the data center. From technologies such as virtualization and cloud computing to data center design, colocation, energy efficiency and monitoring, the BrightTALK data center management community provides the most up-to-date and engaging content from industry experts to better your infrastructure and operations. Engage with a community of your peers and industry experts by asking questions, rating presentations and participating in polls during webinars, all while you gain insight that will help you transform your infrastructure into a next generation data center.

Data Center Management

Welcome to the big data and data management community on BrightTALK. Join thousands of data quality engineers, data scientists, database administrators and other professionals to find more information about the hottest topics affecting your data. Subscribe now to learn about efficiently storing, optimizing a complex infrastructure, developing governing policies, ensuring data quality and analyzing data to make better informed decisions. Join the conversation by watching live and on-demand webinars and take the opportunity to interact with top experts and thought leaders in the field.

Big Data and Data Management

As an IT professional, many of the problems you face are multifaceted, complex and don’t lend themselves to simple solutions. The information technology community features useful and free information technology resources. Join to browse thousands of videos and webinars on ITIL best practices, IT security strategy and more presented by leading CTOs, CIOs and other technology experts.

The Ways Machine Learning and AI Can Fail

Presented by

About this talk

More from this channel