Episodes
-
In this episode Melody chats with Luis Serrano, accomplished Machine Learning Engineer, educator, and author about his new book Grokking Machine Learning.
Luis’ mission is to make information about artificial intelligence and machine learning available to every person in the world. In his new book, Grokking Machine Learning, Luis distills the essential information of machine learning and guides readers toward an intuitive understanding of the field. He believes that behind all of the formulas, there is a soul: one that represents a deeper kind of knowledge accessible to anyone with curiosity. If you are someone who has been intimidated by machine learning or have had trouble describing what exactly you do to friends and family, this is the book for you! -
Experimentation is the foundation of machine learning and artificial intelligence model development. As William Blake said, ‘The true method of knowledge is experiment.’
An experiment is a way to answer the question, what happens when? Like what happens when I try this combination of chess moves? While the principles are the same, experimenting in the digital world of ML is very different than in the physical world. As a human being, I am limited in my ability to try out x number of combinations in a day, because I have to sleep and eat and rest and take mental breaks. But a computer never tires. It can crunch numbers all day and all night until it has tried every possible combination. In ML, there is the unique ability to more freely rely on computational power and big data to run thousands of simulations until some consensus is reached.
In this episode, Saurabh and Melody talk to Alegion’s Chief Data Scientist, Cheryl Martin, all about experimentation in ML and how to fail fast to learn faster. -
Missing episodes?
-
The 2010s was the decade that “machine intelligence” made the leap from sci-fi to reality. When hard-coded rules were replaced by data-driven experience. When human cognition was translated into computer language and machines began to think and learn as humans do.
It has been a decade of great technological advancement and there is so much yet to be accomplished. For example, automation poses exciting growth and progress, as well as unprecedented challenges to society at large. How will we adapt to an automated society, not to mention new technologies like self-driving cars and deep fake imagery? Now that we have this technology, what will we do with it? At the beginning of 2020, we thought it would be nice to look back on the 2010s and consider what this new decade might have in store for us.
In this episode, Nikhil, Saurabh, and I cover a huge range of developments from this past decade including quantum bits, the race to 5G, the Cybertruck, Neuralink, the AI Arms race between the US and China, as well as some of the non-headline making research developments like Adam R and Yolo. Stay tuned to find out why Saurabh claims “Autonomy is taking over our lives in every way.” And how a seemingly innocuous Roomba could be the centerpiece of a grand heist. -
NeurIPS is the machine learning research conference of the year. Although it has been around for 33 years, it has quadrupled in the last five, peaking at 13,000 attendees. NeurIPS is mostly attended by academics (PhD candidates and Post Docs), with a good representation of ML practitioners from industry like Apple, Facebook, Google, and Alegion. The purpose of the conference is to foster the exchange of research on Neural Information Processing Systems, a field that benefits from a combined view of biological, physical, mathematical, and computational sciences. The conference also typically publishes several seminal papers that move ML forward.
In this episode, Alegion’s Chief Data Scientist, Cheryl Martin, and Director of Engineering, Brent Schneeman, share the inside scoop on everything from the best talks, papers, and emergent themes to the shift in focus toward diversity and inclusion including workshops for a host of groups including Women in Machine Learning, Black in AI, LatinX in AI, Queer in AI, {Dis} Ability in AI, and Jews in AI. -
The landscape of every industry is shifting towards automation. Whether you are in software, retail, health, robotics, defense, or any other industry, AI and ML will continue to revolutionize your business model. To evolve your competitive advantage, you need a data science team to develop and support the full life-cycle of your automated systems.
In this episode Melody and Nikhil chat with Alegion CTO Chip Ray about the foundational elements of a strong data science department, the mix of skills, horizons, and expectations unique to each team depending on their objective, and why data engineers deserve as much respect and glory as the data scientists. Finally, they bust up some prevailing myths about data science as a discipline and warn against seeking out magical creatures like the so-called “data science unicorn.” -
The main objective of computer vision is to give machines the ability to see and interpret the world. This has proven a much more complex task than initially expected. We take for granted our innate ability to interpret and classify the world around us. We are attempting to do in decades what took evolution millions of years.
In this episode Saurabh, Nikhil, and Melody discuss the emergence of computer vision as a discipline, the differences in the way that humans and computers “see” images, and the math behind the algorithms. Then they look at some examples of how computer vision is employed in everyday life in Snapchat filters, YouTube video buffering, medical diagnosis of x-rays, and the use of geospatial mapping for agriculture. -
Did you know that not all bias in machine learning (ML) is bad? In fact, the concept of bias was first introduced into ML by Tom Mitchell in his 1980 paper, "The need for biases in learning generalizations.” He defines learning as the ability to generalize from past experience in order to deal with new situations that are related to this experience, but not identical to it. Applying what we’ve learned from past experiences to new situations is called an inductive leap and seems to only be possible if we apply certain biases to choose one generalization about a situation over another. By inserting some types of bias in ML architecture, we give algorithms the capacity to make similar inductive leaps.
The first AI Chair of UNESCO John Shawe-Taylor said, “Humans don’t realize how biased they are until AI reproduces the same bias.” He is referring to the most famous type of bias in ML: human cognitive bias that slips into the training data and skews results. Cognitive bias is a systematic error in thinking that affects the decisions and judgments that people make. Melody, Nikhil, and Saurabh discuss several examples of how cognitive bias has negatively affected models and our society from upside down YouTube videos to an utter lack of facial recognition to Amazon’s AI recruitment tool. -
When discussing machine learning development approaches, data scientists often need to ask themselves does this use case apply best for supervised or unsupervised learning? In this episode, we break down the strengths and weaknesses of each approach and discuss various use cases to which each one best applies. Melody explores the notion that supervised learning works much like our education system: there's a teacher "supervising" the learning process. Unsupervised learning, on the other hand, has no correct answers and no teacher. Algorithms are simply fed unlabeled data and left to structure the data in some new, interesting way. Melody, Nikhil, and Saurabh dive into each approach and cite exciting business use cases including autonomous vehicles, Speech2Face, and accelerating ecological research in Serengeti National Park.
https://content.alegion.com/podcast -
Have you heard that “Data is the new oil”? It sounds cool, but what does it mean? Melody, Nikhil, and Saurabh tease out the ideas behind the metaphor and then discuss why Bernard Marr, a reporter for Forbes, wrote: “Data is not the new oil." They end up offering a different, and perhaps more fitting metaphor to describe what’s fueling the 4th industrial revolution: “AI is the new electricity.”
https://content.alegion.com/podcast -
Welcome to our first episode of No BiaS, where we discuss different perspectives on the emerging and ever-shifting terrain of artificial intelligence and machine learning. In future episodes we’ll dive deeper into the nuts and bolts of developing and training models, philosophical issues, and existential concerns. But since this is our first episode we decided to begin with the basics: AI versus ML. We offer definitions and historical background of how they have evolved over the past few decades into the current state. And then we will peek behind the curtain to discuss the future of the industry.
https://content.alegion.com/podcast
