Q&A with Sentient: on Artificial Intelligence, Open Source Machine Learning Model Management Framework, and the Studio.ML Platform

Reposted from SourceForge.net

The meteoric rise of artificial intelligence (AI), particularly machine learning (ML), is significantly disrupting the business landscape and spurring workforce change across different industries such as manufacturing, transportation, retailing, finance, healthcare, advertising, insurance, entertainment, and more. As businesses enter into a new era ruled by data, organizations can now leverage machine learning algorithms to identify key trends and gain actionable insights by analyzing vast reams of data to make faster and smarter business decisions.

With virtually every industry transforming their business models and core processes to take advantage of machine learning tools, companies should invest in a unifying framework that can help them solve data problems and lessen overhead involved with running, scheduling, monitoring, and managing their machine learning experiments. One cutting-edge open source machine learning model management platform that empowers today’s businesses and data scientists to accelerate and simplify their ML experiments is Studio.ML. Developed by Sentient Technologies, a trusted provider of artificial intelligence solutions, Studio.ML supports data scientists by helping them build on top of their own models and machine learning experiments in a seamless and cost-efficient manner.

SourceForge recently spoke with Arshak Navruzyan, the Chief Technology Officer at Sentient, to shed light on artificial intelligence and open source machine learning. Navruzyan also discusses the benefits of using machine learning model management frameworks and shares how Studio.ML can support data scientists and DevOps to accelerate their scientific research and produce models that help analyze complex and large volumes of data in order to deliver faster, more accurate results.

 

Q: Can you share with us a brief overview of your company? How and when did the company get its start? Who are the brains behind Studio.ML?

A: Sentient was founded in 2007 by Antoine Blondeau (now our Chairman) and Babak Hodjat, our CEO. Together they figured out how to apply Evolutionary Algorithms at massive scale (think 5K GPU, 2M CPUs) and then use this power to create autonomous traders. After proving out their AI platform that traded stocks and shares, they decided to expand beyond finance, as their platform could also be applied to other time-series type problems. As a result, Sentient Technologies was born.

sentient arshak navruzyan

Arshak Navruzyan, Chief Technology Officer at Sentient

Today, we now use our massively scaled AI in two other industries. First, digital marketing, through a product called Sentient Ascend, which provides a new approach to conversion rate optimization (CRO) using genetic algorithms to automate and accelerate the testing of thousands of web page combinations in the same time as a single A/B test. Second, in retail with a product called Sentient Aware, which provides AI-powered product recommendations using deep learning and online learning to curate products for each customer based on their individual, in-the-moment visual preferences.

Q: Who are some of Sentient’s current customers?

A: We have a number of customers across our digital marketing and retail products. The ones that have been publicly announced include Skechers, Rakuten Marketing, Euroflorist, Clicksco, Cosabella, Nexway, and TransWorld Entertainment.

Q: In your opinion, what benefits do artificial intelligence (AI) platforms offer to today’s businesses?

A: AI platforms offer businesses many advantages, from being able to scale their operations by streamlining and adding efficiencies to common tasks, to achieving results that would simply be impossible for humans to achieve. We’re seeing AI take a bigger role in creating preemptive solutions and insights, and as a result, society will become much more efficient as a whole. Think logistics, e-commerce, healthcare, and finance. All these domains and others are starting to see massive gains from AI. We’ll be able to leverage AI systems to help get things to where they need to go faster and cheaper, we’ll be able to enable people to buy things they weren’t even aware existed or even knew they wanted, we’ll be able to help predict fatal diseases before they get past the point of no return. AI is capable of many things, businesses simply need to ask themselves what problems they need to solve and then ask how AI might provide a solution.

Q: In the era of artificial intelligence (AI), what role does open source machine learning technologies play?

A: Open source is playing a critical role in the advancement and adoption of AI. I would argue that it is one of the most crucial underpinnings for the success of AI and why most of the major technology companies have some form of open source offering, including Google’s TensorFlow, Amazon’s MXNet , Facebook’s PyTorch, and Microsoft’s Cognitive Toolkit (CNTK). With all open source projects, having many eyes on the code helps quicker error detection and allows for faster adoption in terms of the innovative applications that can be created. There is also an inbuilt community happy to help out or answer common problems. When it comes specifically to ML, individual data scientist through to large corporations can benefit immediately by having access to proven models and can build on these and then share back for the benefit of the community.

Q: When companies talk about open source machine learning (ML) model management framework, what are they referencing? How can this cutting-edge solution help data scientists and ML professionals lessen the overhead involved with scheduling, running, monitoring, and managing artifacts of a company’s machine learning experiment?

A: Even the best scientists in the field rarely get the models exactly right from the first attempt, and for every new problem there is a trial-and-error phase. The role of the ML model management frameworks is, on one hand, to partially automate the trial-and-error via so-called hyperparameter optimization or search (in essence, search for what training options work best for a given problem), and on the other hand, to keep track of what has been tried and help researchers reuse those results (both their own and the colleagues’ results) – either in a form of model artifacts that can be pushed into production, or, e.g. as learning curves that illuminate areas of improvement.

Q: Data scientists are facing multiple challenges relating to the usage of large datasets and huge amounts of custom hardware and computing resources. How is Sentient helping data scientists and ML professionals solve these issues?

A: Cutting-edge AI and ML research requires cutting-edge computational resources. Because of that, data scientists often have to do a lot of DevOps heavy lifting (setting up instances in the cloud or building a cluster, configuring dependencies, transferring the data) just to reproduce results from a research paper. The open-source framework Studio.ML enables data scientists with ways to reproduce cutting-edge research results using publically available cloud compute providers (Amazon EC2, Google Cloud Computer, Microsoft Azure ) with a single command. This way data scientists can focus on the analytical aspect of their experiments and not spend valuable time doing DevOps.

Q: Let’s talk about your open source model management tool. In what ways can Studio.ML help businesses, academics, and professionals accelerate and simplify their machine learning (ML) experiments?

A: Studio.ML can run any python code seamlessly while providing substantial value in managing, tracking, reproducing, and sharing the experiments. As such, whenever a researcher wants to look at the results or artifacts of the experiment that has been run months ago, there is no need to play archeologist and start the long-forgotten instances in the cloud.

Moreover, the experiment can be reproduced with a single command and shared with a link to central experiment storage. This allows any data scientist to focus on what is truly new and important and not waste time either learning an unfamiliar framework or digging through the old results.

Q: What are the key features and capabilities of your open source model management tool? What makes Studio. ML unique from other AI platforms available in the market?

A: While companies like Facebook, Google, and other large AI research organizations may have similar proprietary tools, there really isn’t anything like Studio.ML available in the open source community. With Studio.ML you can capture experiment information such as Python environment, files, dependencies and logs without modifying the experiment code. You can monitor and organize experiments using a web dashboard that integrates with TensorBoard. Studio.ML will allow you to run experiments locally, remotely, or in the cloud using Google Cloud or Amazon EC2. It can manage artifacts, perform hyperparameter search and create customizable Python environments for remote execution. You also have access the model library, where you can reuse models that have already been created. Another nice feature is that it lets you bid on Amazon spot instances and remotely turn off your machine when it’s done. Studio.ML can serve the resulting models as well, making integration with the production environment a breeze.

Q: Looking ahead, what trends or new technologies do you believe with change and impact the artificial intelligence and machine learning industries? How is Sentient addressing these head on?

A: We believe that the biggest impact in AI will be made by automated approaches to model design (so-called AI creating AI – approaches such as reinforcement learning of neural network architecture or neuroevolution) and by aggregation of knowledge from multiple domains and datasets (transfer learning) that should help in the situations when high-quality labelled data is not available or too expensive. For example, when captioning images for visually impaired people, direct training data (images with proper descriptions) is relatively scarce, however, one has a virtually infinite supply of text or images with simple attributes from the internet that transfer learning should be able to tap into.

At Sentient, our research team is heavily focused on both architecture search using neuroevolution (in fact, there have been several recent breakthroughs in that area whereby neuroevolution was able to find seemingly strange neural network architectures that nonetheless performed better than a hand-crafted state of the art), and on transfer learning. By combining both approaches we demonstrated neural networks that learn with only a few examples per class. We also believe that breakthrough is not possible without expending sufficient amount of computing resources. While (fortunately) these are steadily getting cheaper by virtue of Moore’s law, managing an increasing amount of compute resources can generate an increasing amount hassle, so the role of machine learning management frameworks like Studio.ML will increase.

About Sentient Technologies
Headquartered in San Francisco, Sentient Technologies focuses on artificial intelligence and deep learning to solve real-world problems. With Sentient’s cutting-edge artificial intelligence platform, the company aims to transform the way businesses solve their most complex, mission-critical problems by empowering them to make the right decisions faster. Sentient employs over 100 people and specializes in data analysis, machine learning, and artificial intelligence software and technologies.

Posted in

Want to contribute?

If you have any questions, bug reports or feature requests, please don't hesitate to post on our Github Issues page.