Web Master Web Master

National Drive Electric Week 2018 - Naperville/Chicago, IL

EV+Day.jpg

As I am starting my research to pursue an Electric Vehicle (EV) project in 2019, I found an EV event at Naperville, IL. National Drive Electric Week event was held in Naperville (Chicago) today sponsored by the Fox Valley Electric Auto Association (FVEAA). There were lots of commercial EVs including lots of Tesla models. They also provided an opportunity for people to test drive the EVs. What caught my attention was the classic vehicles that were converted to EV. It was a great opportunity to see the modern EVs first hand as well as seeing different converted EVs. Awsome!

Below is the Youtube video highlighting the event. Enjoy!

Read More
Web Master Web Master

SARC Summer Field Day 2018 - Schaumburg Amateur "Ham" Radio Club

ARRL 2018 Summer Field Day took place on 23 - 24 June 2018.  Schaumburg Amateur "Ham" Radio Club participated in the Summer Field Day at Schaumburg, IL.

Field Day is an annual amateur radio exercise over 24 hour time period.  It is a single largest emergency preparedness exercise that is both fun and educational.  Participants include Ham operators and clubs from the US and Canada.  Each CW (morse code communication), SSB (voice communication), and Digital (using computer over radio frequency) contacts receive points toward the contest,

SARC members did a great job with the final score of 3,204.  35 operators participated over 24 hour time period plus setup & tear down.  They operated on 80, 40, 20, 15, and 10 meter frequency bands.

  • 610 CW QSOs (620 confirmed morse code contacts with other participants in US & Canada)

  • 382 Phone QSOs (382 voice communication contacts)

  • 35 Digital QSOs (Yes, we use computers to "chat" over Ham radio frequency)

  • 490 Bonus Points (extra points for using solar power, younger operators etc)

Great Job everyone!

Read More
Web Master Web Master

TensorFlow Chicago Workshop - Building an Object Detection System in Keras

The TensorFlow Chicago meetup held a one day workshop to build an object detection system in TensorFlow and Keras.  It was a great opportunity to go beyond the traditional introductory materials in deep learning and learn how a real-life applications are developed in Keras and TensorFlow. 

Approximately 50 participants were made up of Data Scientists and developers from different companies, graduate and PhD students in CS and Predictive Analytics.  Also it was really interesting that the group represented people from all over the world from Russia to Brazil.

Above: Image detection did a great job in detecting many objects in the class !!

 

TensorFlow Workshop 

Github Link HERE

  • Session 1. Introduction to deep learning workflow

    • Objective: Walk students through the process we'll be going through during the workshop

  • Session 2. Basics of Deep Learning

    • Objective: Understanding the main concepts around image classification (convolutional neural networks, transfer learning)

  • Session 3. Object Detection

    • Objective: A quick tour of the main concepts developed in the last few years in object detection, ending with Mask-RCNN in Keras

      Architectures include: R-CNN Fast R-CNN Faster R-CNN Mask R-CNN One shot (YOLO, SSD)

  • Session 4. Model fine-tuning, part 1

    • Objective: Fine-tune a pre-trained model on a narrower set of images

  • Session 5. Dataset creation

    • Objective: Generate a dataset unique to the room and object being detected

  • Session 6. Model fine-tuning, part 2

    • Objective: Fine-tune a pre-trained model on the dataset generated in Session 5

SERVER: AWS EC2 with GPU was provided to the students for a hands-on lab

Instructor Profile

Garrett Smith is founder and lead developer of Guild AI (https://guild.ai), an open source toolkit to streamline TensorFlow and Keras model development. Garrett is veteran of software and systems development, having built tools and managed operations for CloudBees platform-as-a-service. In recent years he has focused on deep learning applications, applying best practices in systems engineering to neural network development and deployment.

Rajiv Shah is a data scientist with DataRobot and an Adjunct Assistant Professor at the University of Illinois at Chicago. He has previously worked as a data scientist for State Farm and Caterpillar. He is an active member of the data science community in Chicago and helps organize the Tensorflow Deep Learning meet up. He has a PhD from the University of Illinois at Urbana Champaign. You can find him on twitter at rajcs4 or his home page http://www.rajivshah.com.

Links to reference material

Read More
Web Master Web Master

Google Cloud 2018 - Cloud OnBoard - Chicago, IL

IMG_7443.png

I attended the Google Training event at the downtown Chicago.  It was a great opportunity to learn about Google's latest Cloud Platform.  It was a 1.5 hour train + bus ride to get to the event, but it was well worth it.  I really enjoyed the modules and a big "Thank You" to all the Googlers who hosted the event.  I am totally convinced that the "cloud" is the future.

  • Registration

  • Module 1 - Introducing Google Cloud Platform

  • Module 2 - Getting Started with Google Cloud Platform

  • Lunch

  • Module 3 - Virtual Machines in the Cloud

  • Module 4 - Storage in the Cloud

  • Module 5 - Containers in the Cloud

  • Break

  • Module 6 - Applications in the Cloud

  • Module 7 - Developing, Deploying, and Monitoring in the Cloud

  • Module 8 - Big Data and Machine Learning in the Cloud

Link to the slides here

IMG_7458b.jpg
Read More
Web Master Web Master

Pydata Chicago - Work Hard Once - Python Automation

pic1.jpg

I attended the Pydata meetup event at Arity (222 W Merchandise Mart Plaza Suite 875, Chicago, IL).  Great speech by Franklin Sarkett and also learned a lot from the people who asked insightful questions.  Appreciate the organizers and Arity for hosting a great event.  Everyone enjoyed the great food and the open and innovative environment (In 2016, Allstate spins out a new startup focused on flagging risky drivers).

https://www.meetup.com/PyDataChi/events/249043774/

Abstract: Python is an incredible tool for data science, data engineering, an devops. I would like to present on how to create a tech stack that brings together the different pieces of data science with automation.

Bio: Franklin Sarkett is the cofounder of Audantic Real Estate Analytics. The company predicts home sales for its clients using machine learning algorithms. Previously, Franklin was a data scientist at Facebook and developed an algorithm for the Ads Payments team that increased revenue over $200 million and earned a patent. He has a CS degree from University of Illinois at Urbana Champaign, and a masters in Applied Statistics from DePaul University.

    Pydata Chicago - work hard once

    1. Work Hard Once Strategy and Automation applied to building machine learning models Franklin Sarkett April 2, 2018
    2. About me: franklin.sarkett@gmail.com Audantic Real Estate Analytics, co-founder ● http://audantic.com/ ● Audantic provides customized data, analytics, and predictive analytics for machine residential real estate. Facebook ● Data scientist at Facebook and developed an algorithm for the Ads Payments team that increased revenue over $200 million and earned a patent. Education ● CS degree from University of Illinois at Urbana Champaign ● MS in Applied Statistics from DePaul University.
    3. Summary Building machine learning models from data ingestion to productionalization is challenging, with many steps. Of all the steps, feature engineering is the biggest differentiator between models that work and models that do not. Using automation and strategy we can remove some of the most challenging parts, and focus on the area of machine learning that generates the most value: feature engineering.
    4. John Boyd and the OODA Loop The OODA loop is the decision cycle of observe, orient, decide, and act, developed by military strategist and United States Air Force Colonel John Boyd. Boyd applied the concept to the combat operations process. It is now also often applied to understand commercial operations and learning processes. The approach favors agility over raw power in dealing with human opponents in any endeavor. - Wikipedia
    5. Orient (most important) "Orient" is the key to the OODA loop. Since one is conditioned by one's heritage, surrounding culture, existing knowledge and learnings, the mind combines fragments of ideas, information, conjectures, impressions, etc. to generate our orientation. How well your orientation matches the real world is largely a function of how well you observe.
    6. Stages of Machine Learning Feature engineering Data cleaning Model training Observe Get raw data (sql, csv, API) Orient Decide Model evaluation Deployment Act
    7. Two guiding thoughts A mentor of mine at FB was coaching me on our model building. Building models requires domain knowledge, and put as much data into the model as you can. To improve the models, you need to add: ● Data quality ● Data volume ○ Breadth ○ Depth Addressing these concerns takes Feature Engineering to the next level.
    8. Automating the Observe stage Many of the tasks in the observe stage could be classified as DevOps and Data Engineering. My favorite tools to use for data science: ● Docker ● Jenkins ● Luigi
    9. Orient - Feature Engineering “Coming up with features is difficult, time-consuming, requires expert knowledge. 'Applied machine learning' is basically feature engineering.” — Prof. Andrew Ng.
    10. Orient - Feature Engineering “The algorithms we used are very standard for Kagglers. …We spent most of our efforts in feature engineering. … We were also very careful to discard features likely to expose us to the risk of over-fitting our model.” — Xavier Conort
    11. Orient - Feature Engineering “Feature engineering is the process of transforming raw data into features that better represent the underlying problem to the predictive models, resulting in improved model accuracy on unseen data.” — Dr. Jason Brownlee
    12. Orient - Feature Engineering At the end of the day, some machine learning projects succeed and some fail. What makes the difference? Easily the most important factor is the features used...It is often also one of the most interesting parts, where intuition, creativity and “black art” are as important as the technical stuff. -Pedro Domingos, Prof of CS as University of Washington
    13. Code snippet http://bit.ly/PyDataChi-FeatureEngineering
    14. How do we iterate feature engineering faster? ● Create a pipeline of transforms with a final estimator. ● Pipeline can be used to chain multiple estimators into one. This is useful as there is often a fixed sequence of steps in processing the data, for example feature selection, normalization and classification. ● Benefits: ○ Convenience and encapsulation. You only have to call fit and predict once on your data to fit a whole sequence of estimators ○ Safety. Pipelines help avoid leaking statistics from your test data into the trained model in cross-validation, by ensuring that the same samples are used to train the transformers and predictors.
    15. Feature extraction
    16. Feature extraction
    17. Feature extraction
    18. The pipeline
    19. Summary Building machine learning models from data ingestion to productionalization is hard. Using automation and strategy we can remove some of the most challenging parts, and focus on the area of machine learning that generates the most value: feature engineering. When we use automation and strategy to remove the most challenging parts of machine learning, we can run through more OODA loops faster, generate better models, learn more about our subject, and deliver more value.
    20. franklin.sarkett@gmail.com
    Read More
    Web Master Web Master

    SARC Antenna Tower Project 2018 - Schaumburg Amateur "Ham" Radio Club

    IMG_7022.JPG

    Ray, WA9BLP, is a long time Amateur Ham Radio Operator.  He was having issue with his stations and needed help with his antenna tower.  He has a very tall tower which have been installed over 40 years ago.  It was determined that all coax cables were damaged due to age.  On March 10, 2018, SARC (Schaumburg Amateur Radio Club) volunteers got together to help fix the tower.

     

    Project Summary

    The project stated around 8am and lasted for about 12+ hours.  The weather was clear but the temperature was a bit cold at around 34 to 45 degrees throughout the day.  Approximately 12 volunteers were involved.

    • Replace 4 Feedlines with new 8U COAX and a Rotor cable
    • 1x 3 Element Multi band Antenna
    • 1x Rotatable Dipole (Multi Band)
    • 1x 2 Meter antenna
    • 1x Fixed Dipole antenna
    • Rotor cable
    • Terminate cable with PL259 outside and solder to connection plate in shack, incl. Rotor cable
    • Replace SO239 connectors in shack
    • Replace Matching network on 3 element Multi Band Antenna
    • Visual Check of Rotor, replace junction box for rotor cable connection next to Rotor
    • Drape 5x cables on tower, feed cables into attic then down wall into shack, with drip loop, extra in 2 feet  in attic
    • Verify operation when all ends terminated and tower raised again.
    • Check AMP and AMP power supply, both replaced with spare. (see detail write up from Rob, below)
    • Clean up yard and putt away ladders and supplies

    It is not common for the us to get a real hands on with a large tower.  It was a hard work, but we all learned a lot.  Nothing beats learning by doing.  Also Ray is now back on the air!

    Read More