Posted by David Dooling, RS21 Senior Data Scientist
This is the third installment of a three-part series on how to leverage Artificial Intelligence (AI) to build mission-directed insights for government agencies.
If you haven’t yet, read the previous articles in the series: Part 1 — Natural Language Processing and Part 2 — Data Mining.
With the explosion of data around us today, it can be difficult to pick out the important information from the noise. We often mistakenly deem information irrelevant, but upon further inspection, we find that the signal is clearly advising a course of action.
Signals are comprised of many components, and if a person is only exposed to part of the signal instead of the entire signal, there is a high risk of drawing incorrect conclusions.
For example, imagine that a lifelong fan has season tickets to the local university’s basketball team. She attends every home game, and her team wins them all. Because she only sees the home games but misses the away games, she incorrectly concludes that the team is undefeated. She is unaware that while her team won 20 home games, they lost 5 away games.
Unwittingly, observing a filtered signal like this can lead to false conclusions.
A key part of RS21’s mission is to make meaningful signals more apparent. We connect disparate data sources, analyze them, and visualize findings through intuitive interfaces to enrich understandings and provide actionable information our clients can use to inform decisions.
So how does signal processing support better data analytics and better decision-making?
Human Perception
We receive a constant stream of information from the external environment and react to information we judge to be important. A motorist, for example, will coast to a stop when a traffic light changes from green to red. Drivers are accustomed to this signal and agree to stop as a means to avoid accidents.
But not all information is as easily perceived as the traffic light. In fact, it is estimated that the human brain processes 400 billion bits of information per second, but our conscious minds are only aware of 2,000 of those bits.
We miss the vast amount of information raining down on our five senses (i.e., sight, hearing, taste, smell, and touch). Furthermore, our brains often incorrectly process the information that we do catch.
Consider the static image above. Most people incorrectly perceive the wheels as moving. But the wheels are, in fact, static. You can download it or print a hardcopy and still perceive it to be moving. Why? Because the information encoded in the pixels — the color and intensity — varies across the image, causing asymmetric luminance.
Such optical illusions are commonplace in much of our thinking and how we naively interpret the billions of signals around us. In data science, we make an intentional effort to systematically oppose our inherent cognitive illusions.
The Monty Hall Problem
A signal can be defined as any information that varies in time or space, or both time and space. The traffic light switching from green to red encodes the instructions “STOP”. It is both a signal and an agreed upon set of instructions for motorists to stop and wait for the light to turn green again. While the traffic signal encodes a social norm, many signals are not codified in this way and can create competitive advantages for those who know how to interpret them.
For example, the Monty Hall problem, from the TV game show Let’s Make a Deal, illustrates the mismatch between helpful information signaling us to action and our human intuition that runs counter to its message.
In this problem, a contestant is shown three doors and asked to pick the door he or she believes contains a prize. The host then proceeds to open one of the other doors, revealing that there is no prize behind it. The contestant is given the option of either staying with his or her original choice or switching to the door that the host didn’t open. The question is, does the reveal provide enough information for the contestant to know whether to stay with the original choice or switch?
Most people assume it doesn’t matter. But if you play this game repeatedly, the contestants that switch win the prize twice as often as those who don’t. While the extra information doesn’t always motivate contestants to switch doors, in data science we can use such counter-intuitive insights to better predict patterns and outcomes in various circumstances.
Signal Processing Applications in Government
National Oceanic and Atmospheric Administration (NOAA)
A familiar signal is time series data, such as temperatures recorded at a set of geographically distinct weather stations. Such data can show the passage of Earth orbiting the sun, as represented by periodic high annual temperatures in summer months and low annual temperatures in the winter. Similarly, daily and nightly highs and lows indicate the Earth’s 24-hour rotation.
While this is a simple use of signal processing, there are larger implications for the use of climate data and modeling catastrophic weather events that might necessitate a disaster preparedness response.
Signal and image processing techniques and improved algorithms can help the NOAA and other weather forecasting entities predict and assess the impact of natural disasters: hurricanes, flooding, tsunamis, and severe weather events. More accurate predictions coupled with a better understanding of cascading effects and the geographic area affected will allow institutions and disaster response organizations to prepare, provide communications and outreach, and efficiently deploy resources as needed.
National Institutes of Health (NIH)
As the primary U.S. federal agency responsible for biomedical and public health research, the National Institutes of Health (NIH) is a leader in medical research, treatments, and cures. With the power of genomic signal processing (GSP), genetic diseases can be detected and diagnosed early on. This research is advancing precision medicine, an emerging model that tailors therapeutic tools and medicine for individual patients based on their specific genetics. As this field evolves, the reality of precision medicine will help lead to better treatments and improved outcomes for patients.
Department of Defense (DoD)
Signal processing has tremendous implications for the Department of Defense and the intelligence community dealing with military response in complex environments and real-time situations. Imagine, for example, if military analysts could quickly cut through the noise by converting signals from analog to digital and then analyze these signals to detect and respond to adversarial acts.
Furthermore, imagine the enhanced functions of drones and electronic devices built with better visual, physical, and audio signal processing capabilities. Such devices could collect and process intelligence data, react to voice commands, and automatically respond to precise movements and sounds in the environment. This would ultimately result in better support for military personnel.
Economic + Financial Institutions
Cultural and economic behaviors generate signals. Public equity markets were transformed in the early 1980s by statistical analysts who recognized an enormous advantage for those who could couple company data with stock prices.
For example, two companies in the same business sector, such as Coca-Cola and Pepsi, should have similar stock prices. Any large observed differences permit a profitable trade by betting on the under-performing stock and betting against the over-performing stock. This technique has proven to be very profitable to those who collect and appropriately analyze this augmented signal.
Today, financial analysis depends on signal processing techniques, allowing our economic and financial institutions to more accurately evaluate and respond to short-term and long-term forecasts.
Parting Thoughts
Signal processing has proven to be enlightening in many fields, revealing new insights and providing novel actionable information.
RS21 consults with clients and proactively collects a variety of highly relevant data sources. We ensure the information works in concert to address a specific problem so that a clear melody of meaning can be distilled.
Sign up for our newsletter to learn more about RS21’s work and how data science can help you.
RS21 develops interactive data analytics and visualization products.
We blend an advanced computational capability with a network of world-class experts to provide actionable insights to government organizations including:
- Department of Homeland Security (DHS)
- Cybersecurity and Infrastructure Security Agency (CISA)
- Federal Emergency Management Agency (FEMA)
- Transportation Security Administration (TSA)
- United States Coast Guard (USCG)
- United States Agency for International Development (USAID)
- National Laboratories: Argonne National Laboratory, Idaho National Laboratory, Los Alamos National Laboratory, and Sandia National Laboratory
RS21 is a HUBZone Certified Small Business + GSA IT Schedule 70 Company
