• Probability Management – A Cure for the Flaw of Averages

    By Dr. Sam L. Savage, Exe. Director of ProbabilityManagement.org Meetup Hosted by Postgres Conference and Sponsored by AWS. Agenda 5:00 Conference Social Session, Hand passed Snacks & bar & Networking 6:00 Pizza and Salad, sponsored by Amazon 7:00 Presentation 7:10 Live Streaming on this link: https://www.youtube.com/watch?v= *** Market room is very close to where the cocktail hour will be. *** Abstract: Dr. Sam L. Savage, Executive Director of ProbabilityManagement.org, Author of The Flaw of Averages: Why we Underestimate Risk in the Face of Uncertainty, Adjunct Professor of Civil and Environmental Engineering, Stanford University. The discipline of probability management leverages big data to let organizations estimate the chances of good and bad outcomes of all sorts, to cure the Flaw of Averages. See for example how PG&E is applying it to roll up operational risk https://www.informs.org/ORMS-Today/Public-Articles/December-Volume-43-Number-6/Probability-Management-Rolling-up-operational-risk-at-PG-E The Flaw of Averages states that plans based “average” assumptions are wrong on average. That is, the substitution of single number estimates for uncertainties leads to systematic correctable errors. Yet most plans are still crafted around average demands, prices, completion times, etc., rendering them often worse than useless. This is because quantification of uncertainty has required specialized technology and statistical training. SIPs go a long way toward curing this affliction by providing unambiguous representations of uncertainties as arrays of simulated or historical data. 501(c)(3) nonprofit, ProbabilityManagement.org, has developed the open SIPmath standard for communicating these arrays along with metadata. It is compatible with virtually any software platform including native Excel and has broad implications in risk management and regulation. Communities of SIPmath practice have grown up at Chevron, Lockheed Martin, PG&E, within the military and in government finance. Dr. Savage will demonstrate live interactive models in numerous areas of application using Excel files available to all attendees. No prior understating of statistics is assumed, but for those with extensive training in the subject, this presentation will attempt to repair the damage. Speaker Bio Dr. Sam L. Savage is Executive Director of ProbabilityManagement.org, a 501(c)(3) nonprofit devoted to the communication and calculation of uncertainty. The organization has received funding from Chevron, Lockheed Martin, PG&E, and other, and he is joined on the board by Harry Markowitz, Nobel Laureate in Economics. Dr. Savage is author of "The Flaw of Averages: Why We Underestimate Risk in the Face of Uncertainty" (John WIley & Sons, 2009, 2012), and is an Adjunct Professor in Civil and Environmental Engineering at Stanford University as well as a Fellow of the Judge Business School at Cambridge University. He is the inventor of the Stochastic Information Packet (SIP), an audit-able data array for uncertainty. Dr. Savage received his Ph.D. in computational complexity from Yale University.

  • Introduction of PyTorch and Transfer Learning

    Walmart Labs

    Ravi Ilango, Sr Data Scientist, FogHorn Systems This presentation will be live streamed. Agenda 6:30 Doors Open, Food & Networking 7:00 Presentation *** Please arrive by 7:00 PM due to Security *** Live streaming at 7:10pm: Saturday 8 hours workshop TICKETS: http://bit.ly/sfbay-acm-fall19 Membership discount code: “MEMBER25OFF” Event Details PyTorch is the fastest growing framework to build deep learning algorithms. In this one hour seminar, we will cover the state of the art of deep learning. We will provide an intuitive understanding of model development in PyTorch. To solve real-world problems , we will introduce Transfer Learning, where you can build models on top of those created by Google and others in upcoming workshop on September 28th, 2019. If you are looking to expand your skill set in AI with the latest tools and techniques, this is a workshop you do not want to miss. Come to take a preview of the topic on September 23, 2019. Speakers Bio https://www.linkedin.com/in/raviilango/ Ravi is a Sr Data Scientist at FogHorn Systems, working on a variety of revenue-generating projects for clients involving machine learning and deep learning. He has prior experience as a Sr Data Scientist at Apple for 10 years, and a Sr Program Manager at Applied Materials, among other things. He has an MBA from Santa Clara University, Aeronautics and Production Engineering degree from IIT, Madras, and a number of recent Stanford University ML and AI certificates. (Ravi will be presenting in person, and giving a demo). FogHorn Systems is hiring a data scientist NOW https://www.foghorn.io/data-scientist-2/ We are currently headquartered in Sunnyvale for larger office. To apply, send email to [masked], mentioning in email subject "saw DS job at ACM event". Talk to the hiring manager at the meetup, www.LinkedIn.com/in/GregMakowski.

  • Deep Learning with PyTorch and Transfer Learning - AI Workshop II

    8 HR CLASS - Deep Learning with PyTorch and Transfer Learning TICKETS PURCHASE THROUGH EVENTBRITE: https://www.eventbrite.com/e/deep-learning-with-pytorch-and-transfer-learning-ai-workshop-by-sfbay-acm-tickets-64856725211 Optional short link http://bit.ly/sfbay-acm-fall19 Membership discount code: “MEMBER25OFF” Abstract PyTorch is the fastest growing framework to build deep learning algorithms. In this full-day workshop, we will cover the foundational elements of PyTorch and provide an intuitive understanding of model development from scratch. To solve real-world problems, we will cover a very critical area of AI called Transfer Learning, where you can build models on top of those created by Google and others. So if you are looking to expand your skill set in AI with the latest tools and techniques, this is a workshop you do not want to miss. Content: You will have access to all the notebooks, training material to build your own apps. You should be able to directly work on these using Google Colab. For the Workshop itself, we will have AWS instances available for use. Tickets: $150 - Early Bird Registration [until 9/6] $175 - Regular Registration Group rate $130 / person (Contact yshroff "at" g_m_a_i_l ) Key topics covered • Fundamentals and application of ML / DL Tools, techniques with a focus on PyTorch • .Lab - using PyTorch to build and train deep neural networks. Cover image classification • PyTorch deep dive (Convolutional Neural networks, Recurrent Neural networks, Fault detection) • Lab - Build and train advanced detection models (different use cases) • Transfer learning • Lab - Transfer learning • Optimizing your solution for deployment • Lab - OpenVINO, TorchScript You can expect to take-away from the workshop •Theoretical underpinning of Deep Learning technologies • Practical application of DL frameworks to business problems TARGET AUDIENCE would include people who ... • are comfortable in programming • may already work on consulting projects or in some technical business problem solving role. • It is helpful if you have tried Python, Spark and BigDL before. • You are invited to submit a description of your upcoming machine learning projects or vertical. The instructor will review and may try to incorporate some ideas in the class. Through the meetup site, on the left margin, use the [contact] button. We are seeking TA's who know ML to help the audience. TA applicants should contact the instructor in advance. Use the [contact] button on the left, send email, phone, LinkedIn and ML experience). BEFORE THE CLASS, PREPARATIONS & PRE-LOADING: • For all workshops we will use Jupiter notebooks with Python, Spark and BigDL. • For fun, play around with some neural nets at the TensorFlow Playground (http://playground.tensorflow.org). This is a lecture and lab heavy workshop. You're encouraged to attend our Data Science SIG earlier that week to get the fundamental concepts. We will have several TAs on-site to help with the learning process, but expect the class to move at a fairly fast clip! SCHEDULE 8:00 - 8:30 arrive, register, coffee, network 8:30 - 10:00 lecture / lab 15 min break, coffee 10:15 - 11:30 lecture / lab 45 min break for lunch 12:15 -1:45 lecture / lab 15 min break, coffee, small snacks 2:00 - 3:30 lecture / lab 15 min break, coffee, small snacks 3:45 - 5:15 lecture / lab 15 min Q&A Instructors: Ravi Ilango (Data Scientist, FogHorn) Greg Makowski (https://www.linkedin.com/in/gregmakowski/) has been deploying data mining models for 25 years as the "neural net guy" at American Express/Epsilon. He has developed the analytic internals and automation for 6+ enterprise software systems or SaaS systems. His first convolutional neural net was trained in 1991, a Time Delay Neural Net for speech recognition. Vertical experience includes financial services (credit card, retail banking, bond pricing, ACH payments, fraud detection, customer relationship management (mail, phone, email, banner), retail supply chain among others. He always has something to learn from everybody.

  • RACE your FACTs: Making AI work for Enterprises

    PayPal Town Hall

    By Rama Akkiraju, IBM Fellow, Director of IBM’s Watson Division Agenda 6:30 Doors Open, Food & Networking 7:00 Presentation Live Streaming on this link: https://www.youtube.com/watch? *** Please arrive by 7 PM due to Security *** *** Bring PHOTO ID (passport, driver license, etc.) *** Abstract: There is renewed interest among companies these days to implement and deploy AI models in their business processes either to increase automation or to improve human productivity. AI models are making their way as chatbots in customer support scenarios, as doctors' assistants in hospitals, as legal research assistants in the legal domain, as marketing manager assistants in marketing, and as face detection applications in the security domain, just to name a few use cases. Making AI work for enterprises requires a whole new and different set of concerns to be addressed than those for traditional software applications or for consumer-facing AI models such as targeted advertising and product recommendations. These new concerns include robustness (R), accuracy and adaptability (A), continuous learning (C), explainability (E), fairness (F), accountability (A), consistency (C) and transparency (T). In addition, building high quality and scalable AI models requires a specific kind of discipline, methodology, and tools. Data Scientists and practitioners need prescriptive guidance, tools, methods, and best practices on how to procure data, and build, improve and manage their AI models while addressing the concerns mentioned above. In this talk, I will present our best practices for making AI work for enterprises based on our first-hand experience of building scalable AI models for enterprises. Speaker Bio Rama Akkiraju is an IBM Fellow, Master Inventor and IBM Academy Member, and a Director, at IBM’s Watson Division where she leads the AI operations team with a mission to scale AI for Enterprises. Rama also heads the AI mission of enabling natural, personalized and compassionate conversations between computers and humans. Rama has been named by Forbes as one of the ‘Top 20 Women in AI Research’in May 2017, has been featured in ‘A-Team in AI’by Fortune magazine in July 2018 and named ‘Top 10 pioneering women in AI and Machine Learning’ by Enterprise Management 360. In her career, Rama has worked on agent-based decision support systems, electronic market places, and semantic Web services, for which she led a World-Wide-Web (W3C) standard. Rama has co-authored 4 book chapters and over 100 technical papers. Rama has 18 issued patents and 25+ pending. She is the recipient of 3 best paper awards in AI and Operations Research. Rama holds a Masters degree in Computer Science and has received a gold medal from New York University for her MBA for highest academic excellence. Rama served as the President for ISSIP, a Service Science professional society for 2018 and continues to actively drive AI projects through this professional society.

  • Practical Time-Series Forecast and Anomaly Detection in Python

    By Dr. Ahmed Abdulaal, Data Scientist, eBay. Agenda 6:30 Doors Open, Food & Networking 7:00 Presentation Live Streaming on this link: https://www.youtube.com/watch?v= *** Please arrive by 7 PM due to Security *** *** Bring PHOTO ID (passport, driver license, etc.) *** We will walk through tackling a real-world time-series problem with code in python. First, we shall briefly go over some of the different approaches to tackling general time-series problems from statistical, Bayesian, and machine learning viewpoints with example code. Then, we will discuss the nature of outliers (Anomalies) and the challenges to identifying them in real-world applications, such as e-commerce. Depending on the discussion pace and as time permits, we will demonstrate the data-scientist process of feature engineering, model selection, tuning, ensembling, and evaluation for addressing the specific problem and its challenges. Finally, we’ll go over model-productionalizing and setting up dashboards for results communication for the upcoming longer training class to finish. The objective is to provide a quick overview of the methods and tools for time-series modeling with practical code. This talk acts as a preparatory tutorial for a longer ACM seminar of diving deeper into the science and practice of time-series analysis to come. For the example time-series problem, we’ll use an eCommerce data set with the objective of detecting anomalies such as service interruptions and incidents. The talk assumes that the audience have basic knowledge of the Python programming language and data-handling libraries such as Pandas and Numpy, or equivalent libraries in other languages (R, Octave, Matlab, etc.). Further preparation is not required. A list of reading material and other resources relevant to the discussion topic will be provided for upcoming longer ACM seminar. Speaker Bio: Ahmed Abdulaal had joined eBay’s Operations Analytics team as a Data Scientist in Spring 2018. Ahmed’s experience is in Optimization, Simulation, and Machine Learning with applications to Time-Series data. In Fall 2017, he received his Ph.D. in Industrial Engineering from the University of Miami, where he had worked with time-series energy data and contributed to research with 7 scientific publications covering topics like Building Energy Optimization, Electric Vehicle Routing, and Electricity Patterns Classification. Before graduation, Ahmed interned at the Walt Disney Company as a Decision Scientist, where he had worked on production-level time-series media forecast models, then he joined Comcast NBCUniversal’s Golf Channel Division as an Operations Research Scientist to work on demand forecast and price optimization models, as well as A/B testing. Ahmed received his B.S. in Mechanical Engineering from Cairo University, Egypt, his M..S. in Industrial Engineering, and his M.B.A. from the University of New Haven, Connecticut.

  • Efficiently Protecting Software Innovations on a Global Scale

    -by Steve Bachmann, Bachmann Law Group Agenda 6:30 Doors Open, Food & Networking 7:00 Presentation *** Please arrive by 7 PM due to Security *** *** Bring PHOTO ID (passport, driver license, etc.) *** Event Details To be competitive, software companies often employ designers and developers in different locations and countries to implement and update different versions of software. With this increasing workforce in other countries, software company competition can also stem from other countries, and can provide different flavors of a software product to different markets. To adapt to this developing global stage of software development, competition, and implementation, software companies should use diverse and efficient strategies to protect their innovations both where and when the protection is appropriate and makes business sense. This presentation coves strategies for efficiently protecting software innovations, both individual and families or related innovations, in the US and foreign countries. Efficient methods for protecting an innovation in different countries, from different perspectives, and utilizing trade secret and patent protection will be discussed. Real life examples will be discussed to demonstrate how to implement decisions to maximize innovation protection in a manner that aligns with the goals and budget of a company. Speaker Bio Steve Bachmann, a Bay Area native, is the founder of Bachmann Law Group PC and specializes in patent and intellectual property matters. For over 18 years, Steve has counseled clients on prosecution of U.S. and foreign patent and trademark applications, implementing trade secret programs, intellectual property (IP) portfolio development and strategy, licensing and technology transfer negotiation and drafting, open source, competitor IP analysis and investigations, and IP related due diligence. Steve has substantial experience in obtaining patent protection in numerous areas of software and hardware.‎ Steve has a worked with start-up and Fortune 500 companies and tailors IP services to each client. http://bachmann-law.com

  • Building an artificial brain with neuro-inspired deep learning

    By Dr. Chen-Ping Yu, Founder and CEO of Phiar Agenda 6:30 Doors Open, Food & Networking 7:00 Presentation Live Streaming on this link: https://www.youtube.com/watch?v= *** Please arrive by 7 PM due to Security *** *** Bring PHOTO ID (passport, driver license, etc.) *** The field of Computer Vision explores how to make machines understand visual data in various ways. Modern Computer Vision started out with optimizing statistical machine learning methods primarily using hand-crafted "features" for tasks such as Object Detection, Segmentation, and Tracking. Another inspiration behind Computer Vision techniques comes from looking into the biological vision systems, for example by mimicking the brain's deep layers of interconnected neurons as computational layers to accomplish the same tasks. AlexNet's win in the 2012 ImageNet competition solidified the practically of what is now called Deep Learning, and nowadays most techniques are based on such Deep Learning techniques. Yet, just how biologically inspired are they? Would better knowledge of biological vision improve Deep Learning models further? And if so, what kind of direction would such a futuristic Neural Network take? In this talk, we will first review the recent progress of computer vision and related deep learning models, then we will go through a brief overview of how the biological visual system works, and discuss about building a more brain-inspired neural network, to explore if we are able to answer the three questions above. Speaker Bio: Dr. Chen-Ping Yu is the founder and CEO of Phiar, a company building the first AI-powered augmented reality smartphone navigation solution for driving. He was previously a postdoctoral fellow at Harvard University, researching neuro-inspired deep learning. Chen-Ping received his Ph.D. from Stony Brook University in Computer Vision and Machine Learning, and his M.S. from Penn State University. Chen-Ping's graduate research includes classical machine learning methods for image and video segmentation, neuro-image processing, computational models of the human visual system, and deep learning-based classification and detection models. Chen-Ping has been an NSF Fellow and the recipient of numerous honors and awards, and has published more than 15 scientific publications at top AI and cognitive science venues.

  • [CALL FOR SPEAKERS for: Security, Social Modeling, Data Science]

    The Association of Computing Machinery http://www.acm.org/about-acm/about-the-acm-organization is the world’s largest computing society, handling Computer Science conferences and publications. The San Francisco Bay Area ACM is a local professional chapter, a non-profit 501c(3), founded in 1957. We hold two meetups a month on (1) General Computing on the 3rd Wednesday of the month, and (2) Data Science SIG, on data mining, deep learning or big data on the 4th Monday of the month. Among these Meetups, we recently emphasis Security & Social Modeling discussions, and We generally have[masked] people attending our talks. See also our YouTube channel (https://www.youtube.com/user/sfbayacm) with OVER 140 past talks. And you can find our Security & Social Modeling talks on YouTube playlist: https://www.youtube.com/playlist?list=PL87GtQd0bfJyVsBgkL-TyNzZhsNOYK2v_ . SEEKING SPEAKERS In general, we are seeking speakers to book in advance. Talks could be like something you would see at a computing conference, an educational subject for experienced computing professionals. It is fine to err on the side of more technical, algorithmic or mathematical. If you would like to submit a talk proposal, please provide the following: * 3 available dates (DS on 4th Monday of the month) or (General Computing on 3rd Wed of the month). We skip December for talks. * speaker name, phone, email, LinkedIn (or picture) * talk title * talk description (include any desired links, related reading) * speaker bio (include any desired links) CALL FOR PRESENTATIONS IN SECURITY & SOCIAL MODELING Sample titles: 1. Efficiency Gap function, insufficiency and complementariness Reference: https://www.theatlantic.com/science/archive/2018/01/efficiency-gap-gerrymandering/551492/ 2. Code of Ethics in Machine Learning 3. Ethic in financial product design 4. Ethic in social data collection 5. Ethic in Patent design 6. Deep learning from Chatbots 7. Dimension reduction in social science domains Available dates for speakers in 2019: General Computing talks on 3rd Wed: 1/15/2020, 4/15/2020 Data Science SIG talks on 4th Monday in general: 1/27/2020. CONTACT US: On the left side of the Meetup page, in the "Organizers:" box, there is a "Contact" button you can use for the submission, use "general computing", "S&S" or "DS SIG" talk at the beginning to propose your talk. SPONSORSHIP OPPORTUNITIES You can also contact me (Greg Makowski) about sponsorship opportunities for our non-profit organization. We are run by unpaid volunteers. If you provide financial sponsorship, sponsor food or the video recording for a night or talk series, we can offer either a) a "thank you for the donation letter with our 501c(3) non-profit tax ID" for your tax deduction b) "thank the sponsor" time to address the event audience during the "upcoming events" period of one of our events (7:00 - 7:10) c) opt-in registration information of the attendees d) "thank the sponsor" branding on the video, posted on our YouTube video channel of our talks e) a banner in our monthly email newsletter to 6,000 opt-in bay area computing professionals or a section of our print newsletter to members only f) make a suggestion and we can see what we can do, constrained by our volunteer effort and non-profit status. Thanks, Liana Ye, Chair, and Greg Makowski, Business Development Lead and Data Science SIG Chair