machine learning biology stanford

Last Updated on January 19, 2023

What’s behind the machine learning hype? In this non-technical course, you’ll learn everything you’ve been too afraid to ask about machine learning. There’s no coding required. Hands-on exercises will help you get past the jargon and learn how this exciting technology powers everything from self-driving cars to your personal Amazon shopping suggestions. How does machine learning work, when can you use it, and what is the difference between AI and machine learning? They’re all covered. Gain skills in this hugely in-demand and influential field, and discover why machine learning is for everyone!

How the modern marketer is embracing automation

Thomas Nuss/April 2021Share

Thomas Nuss is CEO of the German agency, Eprofessional, which has a reputation for developing nimble, automation-heavy marketing solutions for global brands, including Vodafone and L’Oréal.

Among marketers, the term “automation” can send shivers up the spine. For me, these are shivers of excitement. I know how to use automation to drive better, faster results with fewer man-hours for my clients. Some others may hear the word “automation” and experience slight shivers of apprehension.

In today’s world — it’s 2021 after all — modern marketers can’t experience these kinds of shivers any longer. To move forward, reach customers, and deliver results in an accelerating digital environment, it’s time to give automation another serious look.

I understand the hesitancy. You may have been burned in the past by automated solutions and thought, “I will never let this happen again”. In fact, there are multiple psychological studies about “algorithm aversion”. People get so upset when algorithms fail that they cannot “forgive” the algorithm, whereas they would easily forgive a colleague for the exact same mistake. Ultimately, this leads to a myriad of human errors as teams attempt to do everything manually.

At Eprofessional, we’re driven by data and experience, and both tell us that automation drives business results. A BCG study makes the argument simple: companies that use automated techniques to deliver relevant ads and content to consumers reported up to 20% higher sales within a few weeks. Better yet, companies that deployed automated marketing campaigns with human input can grow sales by an additional 15 percentage points.

The next level of automation

Our willingness to deploy additional levels of automation alongside human input gives Eprofessional a competitive advantage and keeps our clients happy. We continue to test and learn, recently piloting a new automation feature in Google Ads with a handful of our clients ahead of the global launch in April. The system gives brands the ability to automatically apply a range of recommendations from the Google Ads recommendations page. Instead of manually approving the recommendations each day — boom! — the system optimises campaigns as soon as the opportunity arises. Currently, 17 types of recommendations are covered, helping time-starved marketers shift their focus from manual optimisation to strategic thinking.

A simple illustration showing three toggle buttons, with one toggle shown as ‘on’ with a green overlaid circle. Text in the image says: “Automatically applying certain recommendations can help time-starved marketers shift their focus from manual optimisat

The personal care company L’Oréal Germany became one of our first clients to opt into the new Google pilot, automatically applying certain campaign recommendations and removing the manual work.

Our Director of Search & Automation, Franziska Jaeger, was pleased with L’Oréal’s mindset: “We hear from a lot of clients who are worried about giving up control over important aspects of campaign optimisation, so it was great to see how L’Oréal was so willing to try something new.”

We immediately saw results. Our agency saved significant time: 25 hours in March 2021, for instance. This allowed us to swap our focus to improve the measurement of customer engagement on L’Oréal’s websites, which is a more strategic project. This ultimately led to significantly better business results.

A simple black-and-white illustration showing cosmetic items, including a lipstick and mascara, alongside a green clock. Text in the image says: “25 hours saved in March 2021”.

Here’s what we’ve learned in the last few months as we’ve incorporated more automation and machine learning into other client campaigns:

Vodafone: Dialling up results

The multinational telecommunications firm Vodafone was facing a key challenge: People were going through the initial steps of signing up for subscription plans, but a high proportion did not complete the process. Both teams sought to address this issue by using Vodafone’s first-party customer journey and attribution data to understand each users’ whole journey. Then we fed that data into automated bidding strategies.

Vodafone changed its optimisation strategy in Google Ads to focus on net revenue, giving more budget to keywords that generated paying subscribers while reducing budget to keywords that led to cancellations. The results were impressive: Vodafone increased new subscribers by 32% and reduced sign-ups that didn’t convert to paying customers.

“The use of automation opened up the possibility of growing subscriptions and increasing overall revenue,” said Vodafone’s Head of Search, Jasmin Lehmann. “This has allowed us to shift resources to strategic projects and innovation”.

A simple black-and-white illustration showing three mobile phones layered on top of one another, with a blue pie-chart overlaid on top indicating 32%. Text in the image says: “32% increase in new subscribers”.

Hapag-Lloyd: Navigating demand

We helped set up and test 67 Search campaigns in 67 countries for this shipping and container transportation company. Our key goal was finding relevant customers around the world, with campaigns running across different time zones, languages, and target groups. Automated bidding helped us process every relevant signal — which would be impossible to achieve manually — and allocate budget accordingly.

First we applied a simple strategy of driving as many clicks to the Hapag-Lloyd website as possible, then refined our approach by switching to a cost per acquisition (CPA) model. Conversions increased by 63%, and CPA dropped by 14% within six months.

This strategy has now been rolled out to twice as many countries. “It would have been impossible for us to manage campaigns across time zones, languages, and target groups manually in 144 countries,” said Hapag-Lloyd’s Director of Global Digital Marketing, Jenny Gruner. “Automation allowed us to do that — and allocate budget accordingly.”

A simple black-and-white illustration showing a containership in the water, with a sun, clouds and birds above. The ship is overlaid with a blue pie-chart indicating 63%. Text in the image says: “63% increase in conversions”.

The argument for automation

Some people feel unnerved when I say it, but machines are simply smarter than humans in many areas of business. Of course, there are still many tasks that only humans can do, or humans do best — but by working in conjunction with machines, we can make our work more productive, efficient, and satisfying.

Here are four key considerations to use automation to its full potential in your business:

1. First-party data comes first

Feed your own proprietary first-party data into your automated systems instead of relying on third-party data. First-party data provides clear, unique insights about your customers that no one else will have. It also helps future-proof your business as third-party cookies no longer meet the privacy bar.

2. Better measurement, better results

Once your automated systems are supplied with the right first-party data, ensure you are using suitable measurement techniques to fully understand how your campaigns and sites are performing. This can include implementing advanced attribution techniques to get a clearer picture of the channels that are driving results. This allows you to continually optimise and redirect resources to the best performing channels.

3. Be budget conscious

Within Google Ads, the recommendations page will often highlight opportunities to ramp up budget on one campaign and trim budget in another area. Be ready to pivot for new opportunities and don’t limit yourself with strict budget rules.

4. Switch focus to the strategic

Divert the human resources freed up by automation towards generating new, innovative ideas and pursuing more strategic opportunities. That’s where our human brain power works best.

Biological networks are powerful resources for the discovery of interactions and emergent properties in biological systems, ranging from single-cell to population level. Network approaches have been used many times to combine and amplify signals from individual genes, and have led to remarkable discoveries in biology, including drug discovery, protein function prediction, disease diagnosis, and precision medicine. Furthermore, these approaches have shown broad utility in uncovering new biology and have contributed to new discoveries in wet laboratory experiments.

Mathematical machinery that is central to these approaches is machine learning on networks. The main challenge in machine learning on networks is to find a way to extract information about interactions between nodes and to incorporate that information into a machine learning model. To extract this information from networks, classic machine learning approaches often rely on summary statistics (e.g., degrees or clustering coefficients) or carefully engineered features to measure local neighborhood structures (e.g., network motifs). These classic approaches can be limited because these hand-engineered features are inflexible, they often do not generalize to networks derived from other organisms, tissues and experimental technologies, and can fail on datasets with low experimental coverage.

Recent years have seen a surge in approaches that automatically learn to encode network structure into low-dimensional representations, using transformation techniques based on deep learning and nonlinear dimensionality reduction. The idea behind these representation learning approaches is to learn a data transformation function that maps nodes to points in a low-dimensional vector space, also termed embeddings. Representation learning methods have revolutionized the state-of-the-art in network science and the goal of this tutorial is to open the door for these methods to computational biology and bioinformatics.

The tutorial investigates techniques for biological network modeling, analytics and optimization:

  • How do we represent genome-wide biological data with networks and what techniques can be used to unify data coming in different formats and from different experimental technologies? How do we represent different types of biological objects, such as genes, diseases and drugs, and how do we represent different relation types using networks?
  • How do we automatically extract information about interactions between molecular components? How to incorporate this information into a machine learning model? What methods can be used to embed individual nodes in a network?
  • Fundamental to biological networks is the principle that genes underlying the same phenotype tend to interact. How do we mathematically encode such principles into a machine learning model?
  • How can we embed larger network structures, such as subgraphs of disease proteins, and entire networks, such as molecular graphs, into a low-dimensional vector space? How to deal with highly multi-relational and multi-layered networks?
  • Genes associated with the same function tend to cluster in the same network neighborhood. How do we use this notion to predict new gene functions?
  • Proteins involved in the same disease tend to interact with each other. How do representation learning methods use this information to detect disease modules? How can we prioritize candidate disease proteins?
  • How can representation learning methods identify cell types in mega-scale single-cell genomic networks? How do we integrate tissue-specific information into these methods?
  • Drugs tend to target proteins that are close to proteins associated with diseases that these drugs treat. How do we develop methods that operationalize this notion and generate hypotheses for drug repurposing?

About the author

Study on Scholarship Today -- Check your eligibility for up to 100% scholarship.

Leave a Comment