More stories

  • in

    A universal system for decoding any type of data sent across a network

    Every piece of data that travels over the internet — from paragraphs in an email to 3D graphics in a virtual reality environment — can be altered by the noise it encounters along the way, such as electromagnetic interference from a microwave or Bluetooth device. The data are coded so that when they arrive at their destination, a decoding algorithm can undo the negative effects of that noise and retrieve the original data.

    Since the 1950s, most error-correcting codes and decoding algorithms have been designed together. Each code had a structure that corresponded with a particular, highly complex decoding algorithm, which often required the use of dedicated hardware.

    Researchers at MIT, Boston University, and Maynooth University in Ireland have now created the first silicon chip that is able to decode any code, regardless of its structure, with maximum accuracy, using a universal decoding algorithm called Guessing Random Additive Noise Decoding (GRAND). By eliminating the need for multiple, computationally complex decoders, GRAND enables increased efficiency that could have applications in augmented and virtual reality, gaming, 5G networks, and connected devices that rely on processing a high volume of data with minimal delay.

    The research at MIT is led by Muriel Médard, the Cecil H. and Ida Green Professor in the Department of Electrical Engineering and Computer Science, and was co-authored by Amit Solomon and Wei Ann, both graduate students at MIT; Rabia Tugce Yazicigil, assistant professor of electrical and computer engineering at Boston University; Arslan Riaz and Vaibhav Bansal, both graduate students at Boston University; Ken R. Duffy, director of the Hamilton Institute at the National University of Ireland at Maynooth; and Kevin Galligan, a Maynooth graduate student. The research will be presented at the European Solid-States Device Research and Circuits Conference next week.

    Focus on noise

    One way to think of these codes is as redundant hashes (in this case, a series of 1s and 0s) added to the end of the original data. The rules for the creation of that hash are stored in a specific codebook.

    As the encoded data travel over a network, they are affected by noise, or energy that disrupts the signal, which is often generated by other electronic devices. When that coded data and the noise that affected them arrive at their destination, the decoding algorithm consults its codebook and uses the structure of the hash to guess what the stored information is.

    Instead, GRAND works by guessing the noise that affected the message, and uses the noise pattern to deduce the original information. GRAND generates a series of noise sequences in the order they are likely to occur, subtracts them from the received data, and checks to see if the resulting codeword is in a codebook.

    While the noise appears random in nature, it has a probabilistic structure that allows the algorithm to guess what it might be.

    “In a way, it is similar to troubleshooting. If someone brings their car into the shop, the mechanic doesn’t start by mapping the entire car to blueprints. Instead, they start by asking, ‘What is the most likely thing to go wrong?’ Maybe it just needs gas. If that doesn’t work, what’s next? Maybe the battery is dead?” Médard says.

    Novel hardware

    The GRAND chip uses a three-tiered structure, starting with the simplest possible solutions in the first stage and working up to longer and more complex noise patterns in the two subsequent stages. Each stage operates independently, which increases the throughput of the system and saves power.

    The device is also designed to switch seamlessly between two codebooks. It contains two static random-access memory chips, one that can crack codewords, while the other loads a new codebook and then switches to decoding without any downtime.

    The researchers tested the GRAND chip and found it could effectively decode any moderate redundancy code up to 128 bits in length, with only about a microsecond of latency.

    Médard and her collaborators had previously demonstrated the success of the algorithm, but this new work showcases the effectiveness and efficiency of GRAND in hardware for the first time.

    Developing hardware for the novel decoding algorithm required the researchers to first toss aside their preconceived notions, Médard says.

    “We couldn’t go out and reuse things that had already been done. This was like a complete whiteboard. We had to really think about every single component from scratch. It was a journey of reconsideration. And I think when we do our next chip, there will be things with this first chip that we’ll realize we did out of habit or assumption that we can do better,” she says.

    A chip for the future

    Since GRAND only uses codebooks for verification, the chip not only works with legacy codes but could also be used with codes that haven’t even been introduced yet.

    In the lead-up to 5G implementation, regulators and communications companies struggled to find consensus as to which codes should be used in the new network. Regulators ultimately chose to use two types of traditional codes for 5G infrastructure in different situations. Using GRAND could eliminate the need for that rigid standardization in the future, Médard says.

    The GRAND chip could even open the field of coding to a wave of innovation.

    “For reasons I’m not quite sure of, people approach coding with awe, like it is black magic. The process is mathematically nasty, so people just use codes that already exist. I’m hoping this will recast the discussion so it is not so standards-oriented, enabling people to use codes that already exist and create new codes,” she says.

    Moving forward, Médard and her collaborators plan to tackle the problem of soft detection with a retooled version of the GRAND chip. In soft detection, the received data are less precise.

    They also plan to test the ability of GRAND to crack longer, more complex codes and adjust the structure of the silicon chip to improve its energy efficiency.

    The research was funded by the Battelle Memorial Institute and Science Foundation of Ireland. More

  • in

    Using adversarial attacks to refine molecular energy predictions

    Neural networks (NNs) are increasingly being used to predict new materials, the rate and yield of chemical reactions, and drug-target interactions, among others. For these applications, they are orders of magnitude faster than traditional methods such as quantum mechanical simulations. 

    The price for this agility, however, is reliability. Because machine learning models only interpolate, they may fail when used outside the domain of training data.

    But the part that worried Rafael Gómez-Bombarelli, the Jeffrey Cheah Career Development Professor in the MIT Department of Materials Science and Engineering, and graduate students Daniel Schwalbe-Koda and Aik Rui Tan was that establishing the limits of these machine learning (ML) models is tedious and labor-intensive. 

    This is particularly true for predicting ‘‘potential energy surfaces” (PES), or the map of a molecule’s energy in all its configurations. These surfaces encode the complexities of a molecule into flatlands, valleys, peaks, troughs, and ravines. The most stable configurations of a system are usually in the deep pits — quantum mechanical chasms from which atoms and molecules typically do not escape. 

    In a recent Nature Communications paper, the research team presented a way to demarcate the “safe zone” of a neural network by using “adversarial attacks.” Adversarial attacks have been studied for other classes of problems, such as image classification, but this is the first time that they are being used to sample molecular geometries in a PES. 

    “People have been using uncertainty for active learning for years in ML potentials. The key difference is that they need to run the full ML simulation and evaluate if the NN was reliable, and if it wasn’t, acquire more data, retrain and re-simulate. Meaning that it takes a long time to nail down the right model, and one has to run the ML simulation many times” explains Gómez-Bombarelli.

    The Gómez-Bombarelli lab at MIT works on a synergistic synthesis of first-principles simulation and machine learning that greatly speeds up this process. The actual simulations are run only for a small fraction of these molecules, and all those data are fed into a neural network that learns how to predict the same properties for the rest of the molecules. They have successfully demonstrated these methods for a growing class of novel materials that includes catalysts for producing hydrogen from water, cheaper polymer electrolytes for electric vehicles,  zeolites for molecular sieving, magnetic materials, and more. 

    The challenge, however, is that these neural networks are only as smart as the data they are trained on.  Considering the PES map, 99 percent of the data may fall into one pit, totally missing valleys that are of more interest. 

    Such wrong predictions can have disastrous consequences — think of a self-driving car that fails to identify a person crossing the street.

    One way to find out the uncertainty of a model is to run the same data through multiple versions of it. 

    For this project, the researchers had multiple neural networks predict the potential energy surface from the same data. Where the network is fairly sure of the prediction, the variation between the outputs of different networks is minimal and the surfaces largely converge. When the network is uncertain, the predictions of different models vary widely, producing a range of outputs, any of which could be the correct surface. 

    The spread in the predictions of a “committee of neural networks” is the “uncertainty” at that point. A good model should not just indicate the best prediction, but also indicates the uncertainty about each of these predictions. It’s like the neural network says “this property for material A will have a value of X and I’m highly confident about it.”

    This could have been an elegant solution but for the sheer scale of the combinatorial space. “Each simulation (which is ground feed for the neural network) may take from tens to thousands of CPU hours,” explains Schwalbe-Koda. For the results to be meaningful, multiple models must be run over a sufficient number of points in the PES, an extremely time-consuming process. 

    Instead, the new approach only samples data points from regions of low prediction confidence, corresponding to specific geometries of a molecule. These molecules are then stretched or deformed slightly so that the uncertainty of the neural network committee is maximized. Additional data are computed for these molecules through simulations and then added to the initial training pool. 

    The neural networks are trained again, and a new set of uncertainties are calculated. This process is repeated until the uncertainty associated with various points on the surface becomes well-defined and cannot be decreased any further. 

    Gómez-Bombarelli explains, “We aspire to have a model that is perfect in the regions we care about (i.e., the ones that the simulation will visit) without having had to run the full ML simulation, by making sure that we make it very good in high-likelihood regions where it isn’t.”

    The paper presents several examples of this approach, including predicting complex supramolecular interactions in zeolites. These materials are cavernous crystals that act as molecular sieves with high shape selectivity. They find applications in catalysis, gas separation, and ion exchange, among others.

    Because performing simulations of large zeolite structures is very costly, the researchers show how their method can provide significant savings in computational simulations. They used more than 15,000 examples to train a neural network to predict the potential energy surfaces for these systems. Despite the large cost required to generate the dataset, the final results are mediocre, with only around 80 percent of the neural network-based simulations being successful. To improve the performance of the model using traditional active learning methods, the researchers calculated an additional 5,000 data points, which improved the performance of the neural network potentials to 92 percent.

    However, when the adversarial approach is used to retrain the neural networks, the authors saw a performance jump to 97 percent using only 500 extra points. That’s a remarkable result, the researchers say, especially considering that each of these extra points takes hundreds of CPU hours. 

    This could be the most realistic method to probe the limits of models that researchers use to predict the behavior of materials and the progress of chemical reactions. More

  • in

    Making the case for hydrogen in a zero-carbon economy

    As the United States races to achieve its goal of zero-carbon electricity generation by 2035, energy providers are swiftly ramping up renewable resources such as solar and wind. But because these technologies churn out electrons only when the sun shines and the wind blows, they need backup from other energy sources, especially during seasons of high electric demand. Currently, plants burning fossil fuels, primarily natural gas, fill in the gaps.

    “As we move to more and more renewable penetration, this intermittency will make a greater impact on the electric power system,” says Emre Gençer, a research scientist at the MIT Energy Initiative (MITEI). That’s because grid operators will increasingly resort to fossil-fuel-based “peaker” plants that compensate for the intermittency of the variable renewable energy (VRE) sources of sun and wind. “If we’re to achieve zero-carbon electricity, we must replace all greenhouse gas-emitting sources,” Gençer says.

    Low- and zero-carbon alternatives to greenhouse-gas emitting peaker plants are in development, such as arrays of lithium-ion batteries and hydrogen power generation. But each of these evolving technologies comes with its own set of advantages and constraints, and it has proven difficult to frame the debate about these options in a way that’s useful for policymakers, investors, and utilities engaged in the clean energy transition.

    Now, Gençer and Drake D. Hernandez SM ’21 have come up with a model that makes it possible to pin down the pros and cons of these peaker-plant alternatives with greater precision. Their hybrid technological and economic analysis, based on a detailed inventory of California’s power system, was published online last month in Applied Energy. While their work focuses on the most cost-effective solutions for replacing peaker power plants, it also contains insights intended to contribute to the larger conversation about transforming energy systems.

    “Our study’s essential takeaway is that hydrogen-fired power generation can be the more economical option when compared to lithium-ion batteries — even today, when the costs of hydrogen production, transmission, and storage are very high,” says Hernandez, who worked on the study while a graduate research assistant for MITEI. Adds Gençer, “If there is a place for hydrogen in the cases we analyzed, that suggests there is a promising role for hydrogen to play in the energy transition.”

    Adding up the costs

    California serves as a stellar paradigm for a swiftly shifting power system. The state draws more than 20 percent of its electricity from solar and approximately 7 percent from wind, with more VRE coming online rapidly. This means its peaker plants already play a pivotal role, coming online each evening when the sun goes down or when events such as heat waves drive up electricity use for days at a time.

    “We looked at all the peaker plants in California,” recounts Gençer. “We wanted to know the cost of electricity if we replaced them with hydrogen-fired turbines or with lithium-ion batteries.” The researchers used a core metric called the levelized cost of electricity (LCOE) as a way of comparing the costs of different technologies to each other. LCOE measures the average total cost of building and operating a particular energy-generating asset per unit of total electricity generated over the hypothetical lifetime of that asset.

    Selecting 2019 as their base study year, the team looked at the costs of running natural gas-fired peaker plants, which they defined as plants operating 15 percent of the year in response to gaps in intermittent renewable electricity. In addition, they determined the amount of carbon dioxide released by these plants and the expense of abating these emissions. Much of this information was publicly available.

    Coming up with prices for replacing peaker plants with massive arrays of lithium-ion batteries was also relatively straightforward: “There are no technical limitations to lithium-ion, so you can build as many as you want; but they are super expensive in terms of their footprint for energy storage and the mining required to manufacture them,” says Gençer.

    But then came the hard part: nailing down the costs of hydrogen-fired electricity generation. “The most difficult thing is finding cost assumptions for new technologies,” says Hernandez. “You can’t do this through a literature review, so we had many conversations with equipment manufacturers and plant operators.”

    The team considered two different forms of hydrogen fuel to replace natural gas, one produced through electrolyzer facilities that convert water and electricity into hydrogen, and another that reforms natural gas, yielding hydrogen and carbon waste that can be captured to reduce emissions. They also ran the numbers on retrofitting natural gas plants to burn hydrogen as opposed to building entirely new facilities. Their model includes identification of likely locations throughout the state and expenses involved in constructing these facilities.

    The researchers spent months compiling a giant dataset before setting out on the task of analysis. The results from their modeling were clear: “Hydrogen can be a more cost-effective alternative to lithium-ion batteries for peaking operations on a power grid,” says Hernandez. In addition, notes Gençer, “While certain technologies worked better in particular locations, we found that on average, reforming hydrogen rather than electrolytic hydrogen turned out to be the cheapest option for replacing peaker plants.”

    A tool for energy investors

    When he began this project, Gençer admits he “wasn’t hopeful” about hydrogen replacing natural gas in peaker plants. “It was kind of shocking to see in our different scenarios that there was a place for hydrogen.” That’s because the overall price tag for converting a fossil-fuel based plant to one based on hydrogen is very high, and such conversions likely won’t take place until more sectors of the economy embrace hydrogen, whether as a fuel for transportation or for varied manufacturing and industrial purposes.

    A nascent hydrogen production infrastructure does exist, mainly in the production of ammonia for fertilizer. But enormous investments will be necessary to expand this framework to meet grid-scale needs, driven by purposeful incentives. “With any of the climate solutions proposed today, we will need a carbon tax or carbon pricing; otherwise nobody will switch to new technologies,” says Gençer.

    The researchers believe studies like theirs could help key energy stakeholders make better-informed decisions. To that end, they have integrated their analysis into SESAME, a life cycle and techno-economic assessment tool for a range of energy systems that was developed by MIT researchers. Users can leverage this sophisticated modeling environment to compare costs of energy storage and emissions from different technologies, for instance, or to determine whether it is cost-efficient to replace a natural gas-powered plant with one powered by hydrogen.

    “As utilities, industry, and investors look to decarbonize and achieve zero-emissions targets, they have to weigh the costs of investing in low-carbon technologies today against the potential impacts of climate change moving forward,” says Hernandez, who is currently a senior associate in the energy practice at Charles River Associates. Hydrogen, he believes, will become increasingly cost-competitive as its production costs decline and markets expand.

    A study group member of MITEI’s soon-to-be published Future of Storage study, Gençer knows that hydrogen alone will not usher in a zero-carbon future. But, he says, “Our research shows we need to seriously consider hydrogen in the energy transition, start thinking about key areas where hydrogen should be used, and start making the massive investments necessary.”

    Funding for this research was provided by MITEI’s Low-Carbon Energy Centers and Future of Storage study. More

  • in

    Last-mile routing research challenge awards $175,000 to three winning teams

    Routing is one of the most studied problems in operations research; even small improvements in routing efficiency can save companies money and result in energy savings and reduced environmental impacts. Now, three teams of researchers from universities around the world have received prize money totaling $175,000 for their innovative route optimization models.

    The three teams were the winners of the Amazon Last-Mile Routing Research Challenge, through which the MIT Center for Transportation & Logistics (MIT CTL) and Amazon engaged with a global community of researchers across a range of disciplines, from computer science to business operations to supply chain management, challenging them to build data-driven route optimization models leveraging massive historical route execution data.

    First announced in February, the research challenge attracted more than 2,000 participants from around the world. Two hundred twenty-nine researcher teams formed during the spring to independently develop solutions that incorporated driver know-how into route optimization models with the intent that they would outperform traditional optimization approaches. Out of the 48 teams whose models qualified for the final round of the challenge, three teams’ work stood out above the rest. Amazon provided real operational training data for the models and evaluated submissions, with technical support from MIT CTL scientists.

    In real life, drivers frequently deviate from planned and mathematically optimized route sequences. Drivers carry information about which roads are hard to navigate when traffic is bad, when and where they can easily find parking, which stops can be conveniently served together, and many other factors that existing optimization models simply don’t capture.

    Each model addressed the challenge data in a unique way. The methodological approaches chosen by the participants frequently combined traditional exact and heuristic optimization approaches with nontraditional machine learning methods. On the machine learning side, the most commonly adopted methods were different variants of artificial neural networks, as well as inverse reinforcement learning approaches.

    There were 45 submissions that reached the finalist phase, with team members hailing from 29 countries. Entrants spanned all levels of higher education from final-year undergraduate students to retired faculty. Entries were assessed in a double-blind review process so that the judges would not know what team was attached to each entry.

    The third-place prize of $25,000 was awarded to Okan Arslan and Rasit Abay. Okan is a professor at HEC Montréal, and Rasit is a doctoral student at the University of New South Wales in Australia. The runner-up prize at $50,000 was awarded to MIT’s own Xiaotong Guo, Qingyi Wang, and Baichuan Mo, all doctoral students. The top prize of $100,000 was awarded to Professor William Cook of the University of Waterloo in Canada, Professor Stephan Held of the University of Bonn in Germany, and Professor Emeritus Keld Helsgaun of Roskilde University in Denmark. Congratulations to all winners and contestants were held via webinar on July 30.

    Top-performing teams may be interviewed by Amazon for research roles in the company’s Last Mile organization. MIT CTL will publish and promote short technical papers written by all finalists and might invite top-performing teams to present at MIT. Further, a team led by Matthias Winkenbach, director of the MIT Megacity Logistics Lab, will guest-edit a special issue of Transportation Science, one of the most renowned academic journals in this field, featuring academic papers on topics related to the problem tackled by the research challenge. More

  • in

    Exact symbolic artificial intelligence for faster, better assessment of AI fairness

    The justice system, banks, and private companies use algorithms to make decisions that have profound impacts on people’s lives. Unfortunately, those algorithms are sometimes biased — disproportionately impacting people of color as well as individuals in lower income classes when they apply for loans or jobs, or even when courts decide what bail should be set while a person awaits trial.

    MIT researchers have developed a new artificial intelligence programming language that can assess the fairness of algorithms more exactly, and more quickly, than available alternatives.

    Their Sum-Product Probabilistic Language (SPPL) is a probabilistic programming system. Probabilistic programming is an emerging field at the intersection of programming languages and artificial intelligence that aims to make AI systems much easier to develop, with early successes in computer vision, common-sense data cleaning, and automated data modeling. Probabilistic programming languages make it much easier for programmers to define probabilistic models and carry out probabilistic inference — that is, work backward to infer probable explanations for observed data.

    “There are previous systems that can solve various fairness questions. Our system is not the first; but because our system is specialized and optimized for a certain class of models, it can deliver solutions thousands of times faster,” says Feras Saad, a PhD student in electrical engineering and computer science (EECS) and first author on a recent paper describing the work. Saad adds that the speedups are not insignificant: The system can be up to 3,000 times faster than previous approaches.

    SPPL gives fast, exact solutions to probabilistic inference questions such as “How likely is the model to recommend a loan to someone over age 40?” or “Generate 1,000 synthetic loan applicants, all under age 30, whose loans will be approved.” These inference results are based on SPPL programs that encode probabilistic models of what kinds of applicants are likely, a priori, and also how to classify them. Fairness questions that SPPL can answer include “Is there a difference between the probability of recommending a loan to an immigrant and nonimmigrant applicant with the same socioeconomic status?” or “What’s the probability of a hire, given that the candidate is qualified for the job and from an underrepresented group?”

    SPPL is different from most probabilistic programming languages, as SPPL only allows users to write probabilistic programs for which it can automatically deliver exact probabilistic inference results. SPPL also makes it possible for users to check how fast inference will be, and therefore avoid writing slow programs. In contrast, other probabilistic programming languages such as Gen and Pyro allow users to write down probabilistic programs where the only known ways to do inference are approximate — that is, the results include errors whose nature and magnitude can be hard to characterize.

    Error from approximate probabilistic inference is tolerable in many AI applications. But it is undesirable to have inference errors corrupting results in socially impactful applications of AI, such as automated decision-making, and especially in fairness analysis.

    Jean-Baptiste Tristan, associate professor at Boston College and former research scientist at Oracle Labs, who was not involved in the new research, says, “I’ve worked on fairness analysis in academia and in real-world, large-scale industry settings. SPPL offers improved flexibility and trustworthiness over other PPLs on this challenging and important class of problems due to the expressiveness of the language, its precise and simple semantics, and the speed and soundness of the exact symbolic inference engine.”

    SPPL avoids errors by restricting to a carefully designed class of models that still includes a broad class of AI algorithms, including the decision tree classifiers that are widely used for algorithmic decision-making. SPPL works by compiling probabilistic programs into a specialized data structure called a “sum-product expression.” SPPL further builds on the emerging theme of using probabilistic circuits as a representation that enables efficient probabilistic inference. This approach extends prior work on sum-product networks to models and queries expressed via a probabilistic programming language. However, Saad notes that this approach comes with limitations: “SPPL is substantially faster for analyzing the fairness of a decision tree, for example, but it can’t analyze models like neural networks. Other systems can analyze both neural networks and decision trees, but they tend to be slower and give inexact answers.”

    “SPPL shows that exact probabilistic inference is practical, not just theoretically possible, for a broad class of probabilistic programs,” says Vikash Mansinghka, an MIT principal research scientist and senior author on the paper. “In my lab, we’ve seen symbolic inference driving speed and accuracy improvements in other inference tasks that we previously approached via approximate Monte Carlo and deep learning algorithms. We’ve also been applying SPPL to probabilistic programs learned from real-world databases, to quantify the probability of rare events, generate synthetic proxy data given constraints, and automatically screen data for probable anomalies.”

    The new SPPL probabilistic programming language was presented in June at the ACM SIGPLAN International Conference on Programming Language Design and Implementation (PLDI), in a paper that Saad co-authored with MIT EECS Professor Martin Rinard and Mansinghka. SPPL is implemented in Python and is available open source. More

  • in

    A comprehensive study of technological change

    The societal impacts of technological change can be seen in many domains, from messenger RNA vaccines and automation to drones and climate change. The pace of that technological change can affect its impact, and how quickly a technology improves in performance can be an indicator of its future importance. For decision-makers like investors, entrepreneurs, and policymakers, predicting which technologies are fast improving (and which are overhyped) can mean the difference between success and failure.

    New research from MIT aims to assist in the prediction of technology performance improvement using U.S. patents as a dataset. The study describes 97 percent of the U.S. patent system as a set of 1,757 discrete technology domains, and quantitatively assesses each domain for its improvement potential.

    “The rate of improvement can only be empirically estimated when substantial performance measurements are made over long time periods,” says Anuraag Singh SM ’20, lead author of the paper. “In some large technological fields, including software and clinical medicine, such measures have rarely, if ever, been made.”

    A previous MIT study provided empirical measures for 30 technological domains, but the patent sets identified for those technologies cover less than 15 percent of the patents in the U.S. patent system. The major purpose of this new study is to provide predictions of the performance improvement rates for the thousands of domains not accessed by empirical measurement. To accomplish this, the researchers developed a method using a new probability-based algorithm, machine learning, natural language processing, and patent network analytics.

    Overlap and centrality

    A technology domain, as the researchers define it, consists of sets of artifacts fulfilling a specific function using a specific branch of scientific knowledge. To find the patents that best represent a domain, the team built on previous research conducted by co-author Chris Magee, a professor of the practice of engineering systems within the Institute for Data, Systems, and Society (IDSS). Magee and his colleagues found that by looking for patent overlap between the U.S. and international patent-classification systems, they could quickly identify patents that best represent a technology. The researchers ultimately created a correspondence of all patents within the U.S. patent system to a set of 1,757 technology domains.

    To estimate performance improvement, Singh employed a method refined by co-authors Magee and Giorgio Triulzi, a researcher with the Sociotechnical Systems Research Center (SSRC) within IDSS and an assistant professor at Universidad de los Andes in Colombia. Their method is based on the average “centrality” of patents in the patent citation network. Centrality refers to multiple criteria for determining the ranking or importance of nodes within a network.

    “Our method provides predictions of performance improvement rates for nearly all definable technologies for the first time,” says Singh.

    Those rates vary — from a low of 2 percent per year for the “Mechanical skin treatment — Hair removal and wrinkles” domain to a high of 216 percent per year for the “Dynamic information exchange and support systems integrating multiple channels” domain. The researchers found that most technologies improve slowly; more than 80 percent of technologies improve at less than 25 percent per year. Notably, the number of patents in a technological area was not a strong indicator of a higher improvement rate.

    “Fast-improving domains are concentrated in a few technological areas,” says Magee. “The domains that show improvement rates greater than the predicted rate for integrated chips — 42 percent, from Moore’s law — are predominantly based upon software and algorithms.”

    TechNext Inc.

    The researchers built an online interactive system where domains corresponding to technology-related keywords can be found along with their improvement rates. Users can input a keyword describing a technology and the system returns a prediction of improvement for the technological domain, an automated measure of the quality of the match between the keyword and the domain, and patent sets so that the reader can judge the semantic quality of the match.

    Moving forward, the researchers have founded a new MIT spinoff called TechNext Inc. to further refine this technology and use it to help leaders make better decisions, from budgets to investment priorities to technology policy. Like any inventors, Magee and his colleagues want to protect their intellectual property rights. To that end, they have applied for a patent for their novel system and its unique methodology.

    “Technologies that improve faster win the market,” says Singh. “Our search system enables technology managers, investors, policymakers, and entrepreneurs to quickly look up predictions of improvement rates for specific technologies.”

    Adds Magee: “Our goal is to bring greater accuracy, precision, and repeatability to the as-yet fuzzy art of technology forecasting.” More

  • in

    Study finds lockdowns effective at reducing travel in Sierra Leone

    Throughout the Covid-19 pandemic, governments have used data on people’s movements to inform strategies for containing the spread of the virus. In Europe and the United States, for example, contact-tracing apps have used Bluetooth signals in smartphones to alert people when they’ve spent time near app users who have tested positive for Covid-19. 

    But how can governments make evidence-based decisions in countries where such fine-grained data isn’t available? In recent findings, MIT researchers, in collaboration with Sierra Leone’s government, use cell tower records in Sierra Leone to show that people were traveling less during lockdowns. “When the government implemented novel three-day lockdowns, there was a dual aim to reduce virus spread and also limit social impacts, like increased hunger or food insecurity,” says Professor Lily L. Tsai, MIT Governance Lab’s (MIT GOV/LAB) director and founder. “We wanted to know if shorter lockdowns would be successful.”   

    The research was conducted by MIT GOV/LAB and MIT’s Civic Data Design Lab (CDDL), in partnership with Sierra Leone’s Directorate for Science, Innovation and Technology (DSTI) and Africell, a wireless service provider. The findings will be published as a chapter in the book “Urban Informatics and Future Cities,” a selection of research submitted to the 2021 Computational Urban Planning and Urban Management conference. 

    A proxy for mobility: cell tower records

    Any time someone’s cellphone sends or receives a text, or makes or receives a call, the nearest cell tower is pinged. The tower collects some data (call-detail records, or CDRs), including the date and time of the event and the phone number. By tracking which towers a certain (anonymized) phone number pings, the researchers could approximately measure how much someone was moving around.  

    These measurements showed that, on average, people were traveling less during lockdowns than before lockdowns. Professor Sarah Williams, CDDL’s director, says the analysis also revealed frequently traveled routes, which “allow the government to develop region-specific lockdowns.” 

    While more fine-grained GPS data from smartphones paint a more accurate picture of movement, “there just isn’t a systematic effort in many developing countries to build the infrastructure to collect this data,” says Innocent Ndubuisi-Obi Jr., an MIT GOV/LAB research associate. “In many cases, the closest thing we can use as a proxy for mobility is CDR data.”

    Measuring the effectiveness of lockdowns

    Sierra Leone’s government imposed the three-day lockdown, which required people stay in their homes, in April 2020. A few days after the lockdown ended, a two-week inter-district travel ban began. “Analysis of aggregated CDRs was the quickest means to understanding mobility prior to and during lockdowns,” says Michala Mackay, DSTI’s director and chief operating officer. 

    The data MIT and DSTI received was anonymized — an essential part of ensuring the privacy of the individuals whose data was used. 

    Extracting meaning from the data, though, presented some challenges. Only about 75 percent of adults in Sierra Leone own cellphones, and people sometimes share phones. So the towers pinged by a specific phone might actually represent the movement of several people, and not everyone’s movement will be captured by cell towers. 

    Furthermore, some districts in Sierra Leone have significantly fewer towers than others. When the data were collected, Falaba, a rural district in the northeast, had only five towers, while over 100 towers were clustered in and around Freetown, the capital. In areas with very few towers, it’s harder to detect changes in how much people are traveling. 

    Since each district had a unique tower distribution, the researchers looked at each district separately, establishing a baseline for average distance traveled in each district before the lockdowns, then measuring how movement compared to this average during lockdowns. They found that travel to other districts declined in every district, by as much as 72 percent and by as little as 16 percent. Travel within districts also dropped in all but one district. 

    This map shows change in average distance traveled per trip to other districts in Sierra Leone in 2020.

    Image courtesy of the MIT GOV/LAB and CDDL.

    Previous item
    Next item

    Lockdowns have greater costs in poorer areas

    While movement did decline in all districts, the effect was less dramatic in poorer, more sparsely populated areas. This finding was to be expected; other studies have shown that poorer people often can’t afford to comply with lockdowns, since they can’t take time off work or need to travel to get food. Evidence showing how lockdowns are less effective in poorer areas highlights the importance of distributing resources to poorer areas during crises, which could both provide support during a particularly challenging time and make it less costly for people to comply with social distancing measures. 

    “In low-income communities that demonstrated moderate or low compliance, one of the most common reasons why people left their homes was to search for water,” says Mackay. “A policy takeaway was that lockdowns should only be implemented in extreme cases and for no longer than three days at a time.”

    Throughout the project, the researchers collaborated intimately with DSTI. “This meant government officials learned along with the MIT researchers and added crucial local knowledge,” says Williams. “We hope this model can be replicated elsewhere — especially during crises.” 

    The researchers will be developing an MITx course teaching government officials and MIT students how to collaboratively use CDR data during crises, with a focus on how to do the analysis in a way that protects people’s privacy.

    Ndubuisi-Obi Jr. also has led a training on CDR analysis for Sierra Leonean government officials and has written a guide on how policymakers can use CDRs safely and effectively. “Some of these data sets will help us answer really important policy questions, and we have to balance that with the privacy risks,” he says. More

  • in

    Lockdowns reveal inequities in opportunities for walking activities

    Lockdowns saved lives during the global SARS-CoV-2 pandemic. But as much as they have slowed the spread of Covid-19, there have been some unintended consequences.

    New MIT research shows that lockdowns in 10 metropolitan areas throughout the United States led to a marked reduction in walking. These decreases were mostly seen among residents living in lower-income areas of the city, effectively reducing access to physical activity for minorities and people suffering from illnesses such as obesity and diabetes.

    “Walking is the cheapest, most accessible physical exercise that you can do,” says Esteban Moro, visiting research scientist in the MIT Connection Science Group and senior author on the Nature Communications paper published on June 16. “Places in which people have lower incomes, less park access, and more obesity prevalence were more affected by this walking reduction — which you can think of as another pandemic, the lack of access to affordable exercise.”

    The research focused on recreational versus utilitarian walking done by residents in the U.S. cities of New York, Los Angeles, Chicago, Boston, Miami, Dallas, San Francisco, Seattle, Philadelphia, and Washington D.C. (Utilitarian walking is defined as having a goal; for example, walking to the store or to public transportation. Recreational walking is a walk meant for leisure or exercise.)

    Comparing cellphone data from February 2020 to different time points throughout 2020 lockdowns, the researchers saw an average 70 percent decrease in the number of walks — which remained down by about 18 percent after loosened restrictions — a 50 percent decrease in distance walked, and a 72 percent decrease in utilitarian walking — which remained down by 39 percent even after restrictions were lifted.

    On their face, these findings may not be surprising. When people couldn’t leave their homes, they walked less. But digging deeper into the data yields troubling insights. For example, people in lower-income regions are more likely to rely on public transportation. Lockdowns cut back on those services, meaning fewer people walking to trains and buses.

    Another statistic showed that people in higher-income areas reduced their number of utilitarian walks but were able to replace some of the lost movement with recreational walks around their neighborhoods or in nearby parks.

    “People in higher-income areas generally not only have a park nearby, but also have jobs that give them a degree of flexibility. Jobs that permit them to take a break and walk,” says Moro. “People in the low-income regions often don’t have the ability, the opportunity or even the facilities to actually do this.”

    How it was done

    The researchers used de-identified mobile data obtained through a partnership within the company Cuebiq’s Data for Good COVID-19 Collaborative program. The completely anonymized dataset consisted of GPS locations gathered from smartphone accelerometers from users who opted into the program. Moro and his collaborators took these data and, using specifically designed algorithms, determined when people walked, for how long, and for what purpose. They compared this information from before the pandemic, at different points throughout lockdown, and at a point when most restrictions had been eased. They matched the GPS-identified locations of the smartphones with census data to understand income level and other demographics.

    To make sure their dataset was robust, they only used information from areas that could reasonably be considered pedestrian. The researchers also acknowledge that the dataset may be incomplete, considering people may have occasionally walked without their phones on them.

    Leisure versus utilitarian walks were separated according to distance and/or destination. Utilitarian walks are usually shorter and involve stops at destinations other than the starting point. Leisure walks are longer and usually happen closer to home or in dedicated outdoor spaces.

    For example, many of the walks recorded pre-Covid-19 were short and occurred at around 7 a.m. and between 3 and 5 p.m., which would indicate a walking commute. These bouts of walking were replaced on weekends by short walks around noon.

    The key takeaway is that most walking in cities occurs with the goal of getting to a place. If people don’t have the opportunity to walk to places they need to go, they will reduce their walking activity overall. But when provided opportunity and access, people can supplement utilitarian activity with leisure walking.

    What can be done about it

    Taking into account the public health implications of physical inactivity, the authors argue a reduction in access to walking should be considered a second pandemic and be addressed with the same rigor as the Covid-19 pandemic.

    They suggest several tactical urbanization strategies (defined as non-permanent but easily accessible measures) to increase safety and appeal for both utilitarian and recreational walkers. Many of these have already been implemented in various cities around the world to ease economic and other hardships of the pandemic. Sections of city streets have been closed off to cars on weekends or other non-busy times to allow for pedestrian walking areas. Restaurants have been given curb space to allow for outdoor dining.

    “But most of these pop-up pedestrian areas happen in downtown, where people are high-income and have easier access to more walking opportunities,” notes Moro.

    The same attention needs to be paid to lower-income areas, the researchers argue. This study’s data showed that people explored their own neighborhoods in a recreational way more during lockdown than pre-pandemic. Such wanderings, the researcher say, should be encouraged by making any large, multi-lane intersections safer to cross for the elderly, sick, or those with young children. And local parks, usually seen as places for running laps, should be made more attractive destinations by adding amenities like water fountains, shaded pavilions, and hygiene and sanitation spaces.

    This study was unique in that its data came straight from mobile devices, rather than being self-reported in surveys. This more reliable method of tracking made this study more data-driven than other, similar efforts. And the geotagged data allowed the researchers to dig into socioeconomic trends associated with the findings.

    This is the team’s first analysis of physical activity during and just after lockdown. They hope to use lessons learned from this and planned follow-ups to encourage more permanent adoption of pedestrian-friendly pandemic-era changes.

    The Connection Science Group, co-led by faculty member Alex “Sandy” Pentland — who, along with Moro was a co-author on the paper along with six others from the UK, Brazil, and Australia — is part of the MIT Sociotechnical Systems Research Center within the MIT Institute for Data, Systems, and Society. The collaborative research exemplified in this study is core to the mission of the SSRC; in pairing computer science with public health, the group not only observes trends but also contextualizes data and use them to make improvements for everyone.

    “SSRC merges both the social and technological components of the research,” says Moro. “We’re not only building an analysis, but going beyond that to propose new policies and interventions to change what we are seeing for the better.” More