Entrepreneurship & Technology Industrial Engineering & Operations Research IEOR Emerging Area Professor Award UC Berkeley Ikhlaq Sidhu, UC Berkeley q Chief Scientist and Founder Sutardja Center q Professor in IEOR at UC Berkeley q Created many Berkeley programs q Developed Data-X q Advisor to many firms and executives q Granted over 60 US Patents q Invented technologies used at Skype, HP, US Robotics, IBM, and licensed to many others … q Awarded 3Com’s “Inventor of the Year” q HP – Laser Printer Design q Venture Advisor at Onset Ventures, X-Fund q Numerous Advisory Boards and non-profits All degrees: Electrical Engineering and Computer Science (EECS), BS to Ph.D.
prices to solve Wall Street problem • Prediction applications stock market, sports betting, and more • AI for crime detection, traffic guidance, medical diagnostics, etc. • A version of Zillow that is recalculated with the effects of AirBnB income and many more… IEOR 135 Applied Data Science with Venture Applications Sample Rapid AI/Data Projects
students can’t participate, build, and harness these types of technologies New technologies on the horizon * World is changing * Next Industrial Revolution National and Global Security National Competitiveness The result of skill and behavior mismatch:
key element I’ve seen many technical projects with smart people go off track Why we can’t deliver: • Theoretical understanding without a practical understanding of implementation • Narrow focus: silos of disconnected expertise not leading to any useful work product or innovation • Over-design: way too complex • Not even sure what to create. Wanting implementation specs that no one has. • Expensive cost over-runs on development, sometimes even trying to create something that already exists • Disconnected from technical reality • People not on the same page (misaligned), cannot work with each other, team breakdown.
Story Adaptation Ecosystem, Stakeholders Operational & Financial System Architecture Open Source Tools Components Minimal Implementation Working Model Innovation in Algorithms At Berkeley, we have results: People in our programs can build amazing, working projects in 3 months with a relatively little background in ML, AI, and other data technologies. Applicable to all categories of digital transformation Students/ technical staff Leaders/ Entrepreneur A Solution for Rapid Implementation Innovation Leadership Rapid-X
Vanessa Salas Alexandre Vincent Airfare Data Scraper 14 Final Product Safest Path Suggestion • GREEN: SAFEST PATH • RED: SHORTEST PATH Downtown Berkeley to Cal Memorial Stadium Watch live demo here: https://stayfe.herokuapp.com/ CartilageX: Automated anomaly detection in knee MRIs Iriondo C, Jain D, Muhamedrahimov R, Papanikolaou V, Trotskovsky K, Sun L Commercialization of RecycleAI 1 Image taken of waste object and input into model 2 Model classifies waste object Our Project 3 Object sorted to its appropriate destination - Bin Sorter - Robots - Conveyor Belts Prediction of Bitcoin Prices Aashray Yadav Nicolas Sarquis Bhavya Vashisht Sai Kannan Sampath Mubarak Abdul Kader UC Berkeley | Data-X Berkeley Innovation Index My Dinh Jessica Gu Aaron Lu Dayou Wang Yan Zeng Yujun Zou
1. Deep technical students learn many disconnected theories and skills, but they cannot deliver implementations 2. And they work in teams which cannot deliver innovation within companies, government, and research instiutions
• Numpy, SciPy • Pandas • TensorFlow, Sklearn • SQL to Pandas • NLP / NLTK • Matplotlib Quantitative • Prediction: Regression • ML Classification: Logistic, SVM.. Trees, Forests, Bagging, Boosting,.. • Entropy / Information Topics • Deep Learning examples, including CCNs • Correlations • Markov Processes • LTI Systems: Fourier, Filters where applicable • Control Models where applicable Building Block Code Samples • Webscraping • Stock market live download, simple trading • Convolutional Neural Networks • Next Word Predictor, Spell Checking • Recommendation • Web Crawler • Chatbot, E-mail • Social net interfaces including twitter This class will help you combine math and data concepts The course updates with new tools to stay current. You may learn and use tools not presented in the class project. Often: Working Code First Fill In Theory After
Tools: • Numpy, SciPy • Pandas • TensorFlow, Sklearn • SQL to Pandas • NLP / NLTK • Matplotlib Quantitative • Prediction: Regression • ML Classification: Logistic, SVM.. Trees, Forests, Bagging, Boosting,.. • Entropy / Information Topics • Deep Learning examples, including CCNs • Correlations • Markov Processes • LTI Systems: Fourier, Filters where applicable • Control Models where applicable Building Block Code Samples • Webscraping • Stock market live download, simple trading • Convolutional Neural Networks • Next Word Predictor, Spell Checking • Recommendation • Web Crawler • Chatbot, E-mail • Social net interfaces including twitter Often: Working Code First Fill In Theory After • The ML stack use most commonly used in creating ML/AI/Data applications • Application and systems viewpoint of data and ML • Implementation, architecture, and relevant process to build anything • Statistical, rule based, and hybrid decision systems • Connection with relevant mathematical foundations (entropy, correlation, spectral, LTI, basic prediction, classification) • Practical insight into advanced techniques and tools: (eg. CNNs, NLP, scraping, recurrent networks, etc.) • System modeling for data applications
Download Crawl … Stream or Poll Social Net / IoT Application with Automated Decisions Algorithm Options w/ Tables/Matrix Prediction / Classification Test, train, split Keep state Pandas: Short Term Storage Long Term Storage: SQL and File Formats (JSON, CSV, Excel) Web Possible Output Code Blocks Email Control Decision … Chatbot Feedback from External System (World) Pre- process Natural Language, State Features Blockchain (public ledger or cryptolock) APIs, Services APIs, Services
key element Observation: student projects and professional projects that do well require a different understanding. We created a model and framework to provide these: Rapid-X Model Layers 1. Tools: using vs making. Learn to use and understand state of the art tools and technical approaches 2. Theory: Understanding the theory and frameworks behind the tools using first principals 3. Projects: Story first, second is development agility and stakeholders acquisition 4. Project Viewpoints: 5 Viewpoints integrated into the teaching model 5. Behaviors and Mindsets: 6 Behaviors and mindsets tuned for innovation
key element Notes: # * combinations of on-line active systems, API economy, powerful open source tools, live systems that must run and be current all the time, cloud infrastructure compute and storage blocks, .. Rapid-X Model Layers 5 Project Viewpoints: a) Customer touchpoints b) Systems and architecture c) Risk mitigation, d) Agile increments, e) Swim-lanes and team dynamics 6 Necessary Behaviors and Mindsets: a) The target is moving, b) Tools are powerful - use them c) The system is the whole world * d) There is no greenfield - connect to the existing structure, means know the existing structure e) You can’t know it all before you start f) Develop insight, use technical/theoretical analogies, first principals, but don’t just plug and play