More stories

  • in

    Technique could efficiently solve partial differential equations for numerous applications

    In fields such as physics and engineering, partial differential equations (PDEs) are used to model complex physical processes to generate insight into how some of the most complicated physical and natural systems in the world function.

    To solve these difficult equations, researchers use high-fidelity numerical solvers, which can be very time-consuming and computationally expensive to run. The current simplified alternative, data-driven surrogate models, compute the goal property of a solution to PDEs rather than the whole solution. Those are trained on a set of data that has been generated by the high-fidelity solver, to predict the output of the PDEs for new inputs. This is data-intensive and expensive because complex physical systems require a large number of simulations to generate enough data. 

    In a new paper, “Physics-enhanced deep surrogates for partial differential equations,” published in December in Nature Machine Intelligence, a new method is proposed for developing data-driven surrogate models for complex physical systems in such fields as mechanics, optics, thermal transport, fluid dynamics, physical chemistry, and climate models.

    The paper was authored by MIT’s professor of applied mathematics Steven G. Johnson along with Payel Das and Youssef Mroueh of the MIT-IBM Watson AI Lab and IBM Research; Chris Rackauckas of Julia Lab; and Raphaël Pestourie, a former MIT postdoc who is now at Georgia Tech. The authors call their method “physics-enhanced deep surrogate” (PEDS), which combines a low-fidelity, explainable physics simulator with a neural network generator. The neural network generator is trained end-to-end to match the output of the high-fidelity numerical solver.

    “My aspiration is to replace the inefficient process of trial and error with systematic, computer-aided simulation and optimization,” says Pestourie. “Recent breakthroughs in AI like the large language model of ChatGPT rely on hundreds of billions of parameters and require vast amounts of resources to train and evaluate. In contrast, PEDS is affordable to all because it is incredibly efficient in computing resources and has a very low barrier in terms of infrastructure needed to use it.”

    In the article, they show that PEDS surrogates can be up to three times more accurate than an ensemble of feedforward neural networks with limited data (approximately 1,000 training points), and reduce the training data needed by at least a factor of 100 to achieve a target error of 5 percent. Developed using the MIT-designed Julia programming language, this scientific machine-learning method is thus efficient in both computing and data.

    The authors also report that PEDS provides a general, data-driven strategy to bridge the gap between a vast array of simplified physical models with corresponding brute-force numerical solvers modeling complex systems. This technique offers accuracy, speed, data efficiency, and physical insights into the process.

    Says Pestourie, “Since the 2000s, as computing capabilities improved, the trend of scientific models has been to increase the number of parameters to fit the data better, sometimes at the cost of a lower predictive accuracy. PEDS does the opposite by choosing its parameters smartly. It leverages the technology of automatic differentiation to train a neural network that makes a model with few parameters accurate.”

    “The main challenge that prevents surrogate models from being used more widely in engineering is the curse of dimensionality — the fact that the needed data to train a model increases exponentially with the number of model variables,” says Pestourie. “PEDS reduces this curse by incorporating information from the data and from the field knowledge in the form of a low-fidelity model solver.”

    The researchers say that PEDS has the potential to revive a whole body of the pre-2000 literature dedicated to minimal models — intuitive models that PEDS could make more accurate while also being predictive for surrogate model applications.

    “The application of the PEDS framework is beyond what we showed in this study,” says Das. “Complex physical systems governed by PDEs are ubiquitous, from climate modeling to seismic modeling and beyond. Our physics-inspired fast and explainable surrogate models will be of great use in those applications, and play a complementary role to other emerging techniques, like foundation models.”

    The research was supported by the MIT-IBM Watson AI Lab and the U.S. Army Research Office through the Institute for Soldier Nanotechnologies.  More

  • in

    Leveraging language to understand machines

    Natural language conveys ideas, actions, information, and intent through context and syntax; further, there are volumes of it contained in databases. This makes it an excellent source of data to train machine-learning systems on. Two master’s of engineering students in the 6A MEng Thesis Program at MIT, Irene Terpstra ’23 and Rujul Gandhi ’22, are working with mentors in the MIT-IBM Watson AI Lab to use this power of natural language to build AI systems.

    As computing is becoming more advanced, researchers are looking to improve the hardware that they run on; this means innovating to create new computer chips. And, since there is literature already available on modifications that can be made to achieve certain parameters and performance, Terpstra and her mentors and advisors Anantha Chandrakasan, MIT School of Engineering dean and the Vannevar Bush Professor of Electrical Engineering and Computer Science, and IBM’s researcher Xin Zhang, are developing an AI algorithm that assists in chip design.

    “I’m creating a workflow to systematically analyze how these language models can help the circuit design process. What reasoning powers do they have, and how can it be integrated into the chip design process?” says Terpstra. “And then on the other side, if that proves to be useful enough, [we’ll] see if they can automatically design the chips themselves, attaching it to a reinforcement learning algorithm.”

    To do this, Terpstra’s team is creating an AI system that can iterate on different designs. It means experimenting with various pre-trained large language models (like ChatGPT, Llama 2, and Bard), using an open-source circuit simulator language called NGspice, which has the parameters of the chip in code form, and a reinforcement learning algorithm. With text prompts, researchers will be able to query how the physical chip should be modified to achieve a certain goal in the language model and produced guidance for adjustments. This is then transferred into a reinforcement learning algorithm that updates the circuit design and outputs new physical parameters of the chip.

    “The final goal would be to combine the reasoning powers and the knowledge base that is baked into these large language models and combine that with the optimization power of the reinforcement learning algorithms and have that design the chip itself,” says Terpstra.

    Rujul Gandhi works with the raw language itself. As an undergraduate at MIT, Gandhi explored linguistics and computer sciences, putting them together in her MEng work. “I’ve been interested in communication, both between just humans and between humans and computers,” Gandhi says.

    Robots or other interactive AI systems are one area where communication needs to be understood by both humans and machines. Researchers often write instructions for robots using formal logic. This helps ensure that commands are being followed safely and as intended, but formal logic can be difficult for users to understand, while natural language comes easily. To ensure this smooth communication, Gandhi and her advisors Yang Zhang of IBM and MIT assistant professor Chuchu Fan are building a parser that converts natural language instructions into a machine-friendly form. Leveraging the linguistic structure encoded by the pre-trained encoder-decoder model T5, and a dataset of annotated, basic English commands for performing certain tasks, Gandhi’s system identifies the smallest logical units, or atomic propositions, which are present in a given instruction.

    “Once you’ve given your instruction, the model identifies all the smaller sub-tasks you want it to carry out,” Gandhi says. “Then, using a large language model, each sub-task can be compared against the available actions and objects in the robot’s world, and if any sub-task can’t be carried out because a certain object is not recognized, or an action is not possible, the system can stop right there to ask the user for help.”

    This approach of breaking instructions into sub-tasks also allows her system to understand logical dependencies expressed in English, like, “do task X until event Y happens.” Gandhi uses a dataset of step-by-step instructions across robot task domains like navigation and manipulation, with a focus on household tasks. Using data that are written just the way humans would talk to each other has many advantages, she says, because it means a user can be more flexible about how they phrase their instructions.

    Another of Gandhi’s projects involves developing speech models. In the context of speech recognition, some languages are considered “low resource” since they might not have a lot of transcribed speech available, or might not have a written form at all. “One of the reasons I applied to this internship at the MIT-IBM Watson AI Lab was an interest in language processing for low-resource languages,” she says. “A lot of language models today are very data-driven, and when it’s not that easy to acquire all of that data, that’s when you need to use the limited data efficiently.” 

    Speech is just a stream of sound waves, but humans having a conversation can easily figure out where words and thoughts start and end. In speech processing, both humans and language models use their existing vocabulary to recognize word boundaries and understand the meaning. In low- or no-resource languages, a written vocabulary might not exist at all, so researchers can’t provide one to the model. Instead, the model can make note of what sound sequences occur together more frequently than others, and infer that those might be individual words or concepts. In Gandhi’s research group, these inferred words are then collected into a pseudo-vocabulary that serves as a labeling method for the low-resource language, creating labeled data for further applications.

    The applications for language technology are “pretty much everywhere,” Gandhi says. “You could imagine people being able to interact with software and devices in their native language, their native dialect. You could imagine improving all the voice assistants that we use. You could imagine it being used for translation or interpretation.” More

  • in

    Image recognition accuracy: An unseen challenge confounding today’s AI

    Imagine you are scrolling through the photos on your phone and you come across an image that at first you can’t recognize. It looks like maybe something fuzzy on the couch; could it be a pillow or a coat? After a couple of seconds it clicks — of course! That ball of fluff is your friend’s cat, Mocha. While some of your photos could be understood in an instant, why was this cat photo much more difficult?

    MIT Computer Science and Artificial Intelligence Laboratory (CSAIL) researchers were surprised to find that despite the critical importance of understanding visual data in pivotal areas ranging from health care to transportation to household devices, the notion of an image’s recognition difficulty for humans has been almost entirely ignored. One of the major drivers of progress in deep learning-based AI has been datasets, yet we know little about how data drives progress in large-scale deep learning beyond that bigger is better.

    In real-world applications that require understanding visual data, humans outperform object recognition models despite the fact that models perform well on current datasets, including those explicitly designed to challenge machines with debiased images or distribution shifts. This problem persists, in part, because we have no guidance on the absolute difficulty of an image or dataset. Without controlling for the difficulty of images used for evaluation, it’s hard to objectively assess progress toward human-level performance, to cover the range of human abilities, and to increase the challenge posed by a dataset.

    To fill in this knowledge gap, David Mayo, an MIT PhD student in electrical engineering and computer science and a CSAIL affiliate, delved into the deep world of image datasets, exploring why certain images are more difficult for humans and machines to recognize than others. “Some images inherently take longer to recognize, and it’s essential to understand the brain’s activity during this process and its relation to machine learning models. Perhaps there are complex neural circuits or unique mechanisms missing in our current models, visible only when tested with challenging visual stimuli. This exploration is crucial for comprehending and enhancing machine vision models,” says Mayo, a lead author of a new paper on the work.

    This led to the development of a new metric, the “minimum viewing time” (MVT), which quantifies the difficulty of recognizing an image based on how long a person needs to view it before making a correct identification. Using a subset of ImageNet, a popular dataset in machine learning, and ObjectNet, a dataset designed to test object recognition robustness, the team showed images to participants for varying durations from as short as 17 milliseconds to as long as 10 seconds, and asked them to choose the correct object from a set of 50 options. After over 200,000 image presentation trials, the team found that existing test sets, including ObjectNet, appeared skewed toward easier, shorter MVT images, with the vast majority of benchmark performance derived from images that are easy for humans.

    The project identified interesting trends in model performance — particularly in relation to scaling. Larger models showed considerable improvement on simpler images but made less progress on more challenging images. The CLIP models, which incorporate both language and vision, stood out as they moved in the direction of more human-like recognition.

    “Traditionally, object recognition datasets have been skewed towards less-complex images, a practice that has led to an inflation in model performance metrics, not truly reflective of a model’s robustness or its ability to tackle complex visual tasks. Our research reveals that harder images pose a more acute challenge, causing a distribution shift that is often not accounted for in standard evaluations,” says Mayo. “We released image sets tagged by difficulty along with tools to automatically compute MVT, enabling MVT to be added to existing benchmarks and extended to various applications. These include measuring test set difficulty before deploying real-world systems, discovering neural correlates of image difficulty, and advancing object recognition techniques to close the gap between benchmark and real-world performance.”

    “One of my biggest takeaways is that we now have another dimension to evaluate models on. We want models that are able to recognize any image even if — perhaps especially if — it’s hard for a human to recognize. We’re the first to quantify what this would mean. Our results show that not only is this not the case with today’s state of the art, but also that our current evaluation methods don’t have the ability to tell us when it is the case because standard datasets are so skewed toward easy images,” says Jesse Cummings, an MIT graduate student in electrical engineering and computer science and co-first author with Mayo on the paper.

    From ObjectNet to MVT

    A few years ago, the team behind this project identified a significant challenge in the field of machine learning: Models were struggling with out-of-distribution images, or images that were not well-represented in the training data. Enter ObjectNet, a dataset comprised of images collected from real-life settings. The dataset helped illuminate the performance gap between machine learning models and human recognition abilities, by eliminating spurious correlations present in other benchmarks — for example, between an object and its background. ObjectNet illuminated the gap between the performance of machine vision models on datasets and in real-world applications, encouraging use for many researchers and developers — which subsequently improved model performance.

    Fast forward to the present, and the team has taken their research a step further with MVT. Unlike traditional methods that focus on absolute performance, this new approach assesses how models perform by contrasting their responses to the easiest and hardest images. The study further explored how image difficulty could be explained and tested for similarity to human visual processing. Using metrics like c-score, prediction depth, and adversarial robustness, the team found that harder images are processed differently by networks. “While there are observable trends, such as easier images being more prototypical, a comprehensive semantic explanation of image difficulty continues to elude the scientific community,” says Mayo.

    In the realm of health care, for example, the pertinence of understanding visual complexity becomes even more pronounced. The ability of AI models to interpret medical images, such as X-rays, is subject to the diversity and difficulty distribution of the images. The researchers advocate for a meticulous analysis of difficulty distribution tailored for professionals, ensuring AI systems are evaluated based on expert standards, rather than layperson interpretations.

    Mayo and Cummings are currently looking at neurological underpinnings of visual recognition as well, probing into whether the brain exhibits differential activity when processing easy versus challenging images. The study aims to unravel whether complex images recruit additional brain areas not typically associated with visual processing, hopefully helping demystify how our brains accurately and efficiently decode the visual world.

    Toward human-level performance

    Looking ahead, the researchers are not only focused on exploring ways to enhance AI’s predictive capabilities regarding image difficulty. The team is working on identifying correlations with viewing-time difficulty in order to generate harder or easier versions of images.

    Despite the study’s significant strides, the researchers acknowledge limitations, particularly in terms of the separation of object recognition from visual search tasks. The current methodology does concentrate on recognizing objects, leaving out the complexities introduced by cluttered images.

    “This comprehensive approach addresses the long-standing challenge of objectively assessing progress towards human-level performance in object recognition and opens new avenues for understanding and advancing the field,” says Mayo. “With the potential to adapt the Minimum Viewing Time difficulty metric for a variety of visual tasks, this work paves the way for more robust, human-like performance in object recognition, ensuring that models are truly put to the test and are ready for the complexities of real-world visual understanding.”

    “This is a fascinating study of how human perception can be used to identify weaknesses in the ways AI vision models are typically benchmarked, which overestimate AI performance by concentrating on easy images,” says Alan L. Yuille, Bloomberg Distinguished Professor of Cognitive Science and Computer Science at Johns Hopkins University, who was not involved in the paper. “This will help develop more realistic benchmarks leading not only to improvements to AI but also make fairer comparisons between AI and human perception.” 

    “It’s widely claimed that computer vision systems now outperform humans, and on some benchmark datasets, that’s true,” says Anthropic technical staff member Simon Kornblith PhD ’17, who was also not involved in this work. “However, a lot of the difficulty in those benchmarks comes from the obscurity of what’s in the images; the average person just doesn’t know enough to classify different breeds of dogs. This work instead focuses on images that people can only get right if given enough time. These images are generally much harder for computer vision systems, but the best systems are only a bit worse than humans.”

    Mayo, Cummings, and Xinyu Lin MEng ’22 wrote the paper alongside CSAIL Research Scientist Andrei Barbu, CSAIL Principal Research Scientist Boris Katz, and MIT-IBM Watson AI Lab Principal Researcher Dan Gutfreund. The researchers are affiliates of the MIT Center for Brains, Minds, and Machines.

    The team is presenting their work at the 2023 Conference on Neural Information Processing Systems (NeurIPS). More

  • in

    Technique enables AI on edge devices to keep learning over time

    Personalized deep-learning models can enable artificial intelligence chatbots that adapt to understand a user’s accent or smart keyboards that continuously update to better predict the next word based on someone’s typing history. This customization requires constant fine-tuning of a machine-learning model with new data.

    Because smartphones and other edge devices lack the memory and computational power necessary for this fine-tuning process, user data are typically uploaded to cloud servers where the model is updated. But data transmission uses a great deal of energy, and sending sensitive user data to a cloud server poses a security risk.  

    Researchers from MIT, the MIT-IBM Watson AI Lab, and elsewhere developed a technique that enables deep-learning models to efficiently adapt to new sensor data directly on an edge device.

    Their on-device training method, called PockEngine, determines which parts of a huge machine-learning model need to be updated to improve accuracy, and only stores and computes with those specific pieces. It performs the bulk of these computations while the model is being prepared, before runtime, which minimizes computational overhead and boosts the speed of the fine-tuning process.    

    When compared to other methods, PockEngine significantly sped up on-device training, performing up to 15 times faster on some hardware platforms. Moreover, PockEngine didn’t cause models to have any dip in accuracy. The researchers also found that their fine-tuning method enabled a popular AI chatbot to answer complex questions more accurately.

    “On-device fine-tuning can enable better privacy, lower costs, customization ability, and also lifelong learning, but it is not easy. Everything has to happen with a limited number of resources. We want to be able to run not only inference but also training on an edge device. With PockEngine, now we can,” says Song Han, an associate professor in the Department of Electrical Engineering and Computer Science (EECS), a member of the MIT-IBM Watson AI Lab, a distinguished scientist at NVIDIA, and senior author of an open-access paper describing PockEngine.

    Han is joined on the paper by lead author Ligeng Zhu, an EECS graduate student, as well as others at MIT, the MIT-IBM Watson AI Lab, and the University of California San Diego. The paper was recently presented at the IEEE/ACM International Symposium on Microarchitecture.

    Layer by layer

    Deep-learning models are based on neural networks, which comprise many interconnected layers of nodes, or “neurons,” that process data to make a prediction. When the model is run, a process called inference, a data input (such as an image) is passed from layer to layer until the prediction (perhaps the image label) is output at the end. During inference, each layer no longer needs to be stored after it processes the input.

    But during training and fine-tuning, the model undergoes a process known as backpropagation. In backpropagation, the output is compared to the correct answer, and then the model is run in reverse. Each layer is updated as the model’s output gets closer to the correct answer.

    Because each layer may need to be updated, the entire model and intermediate results must be stored, making fine-tuning more memory demanding than inference

    However, not all layers in the neural network are important for improving accuracy. And even for layers that are important, the entire layer may not need to be updated. Those layers, and pieces of layers, don’t need to be stored. Furthermore, one may not need to go all the way back to the first layer to improve accuracy — the process could be stopped somewhere in the middle.

    PockEngine takes advantage of these factors to speed up the fine-tuning process and cut down on the amount of computation and memory required.

    The system first fine-tunes each layer, one at a time, on a certain task and measures the accuracy improvement after each individual layer. In this way, PockEngine identifies the contribution of each layer, as well as trade-offs between accuracy and fine-tuning cost, and automatically determines the percentage of each layer that needs to be fine-tuned.

    “This method matches the accuracy very well compared to full back propagation on different tasks and different neural networks,” Han adds.

    A pared-down model

    Conventionally, the backpropagation graph is generated during runtime, which involves a great deal of computation. Instead, PockEngine does this during compile time, while the model is being prepared for deployment.

    PockEngine deletes bits of code to remove unnecessary layers or pieces of layers, creating a pared-down graph of the model to be used during runtime. It then performs other optimizations on this graph to further improve efficiency.

    Since all this only needs to be done once, it saves on computational overhead for runtime.

    “It is like before setting out on a hiking trip. At home, you would do careful planning — which trails are you going to go on, which trails are you going to ignore. So then at execution time, when you are actually hiking, you already have a very careful plan to follow,” Han explains.

    When they applied PockEngine to deep-learning models on different edge devices, including Apple M1 Chips and the digital signal processors common in many smartphones and Raspberry Pi computers, it performed on-device training up to 15 times faster, without any drop in accuracy. PockEngine also significantly slashed the amount of memory required for fine-tuning.

    The team also applied the technique to the large language model Llama-V2. With large language models, the fine-tuning process involves providing many examples, and it’s crucial for the model to learn how to interact with users, Han says. The process is also important for models tasked with solving complex problems or reasoning about solutions.

    For instance, Llama-V2 models that were fine-tuned using PockEngine answered the question “What was Michael Jackson’s last album?” correctly, while models that weren’t fine-tuned failed. PockEngine cut the time it took for each iteration of the fine-tuning process from about seven seconds to less than one second on a NVIDIA Jetson Orin, an edge GPU platform.

    In the future, the researchers want to use PockEngine to fine-tune even larger models designed to process text and images together.

    “This work addresses growing efficiency challenges posed by the adoption of large AI models such as LLMs across diverse applications in many different industries. It not only holds promise for edge applications that incorporate larger models, but also for lowering the cost of maintaining and updating large AI models in the cloud,” says Ehry MacRostie, a senior manager in Amazon’s Artificial General Intelligence division who was not involved in this study but works with MIT on related AI research through the MIT-Amazon Science Hub.

    This work was supported, in part, by the MIT-IBM Watson AI Lab, the MIT AI Hardware Program, the MIT-Amazon Science Hub, the National Science Foundation (NSF), and the Qualcomm Innovation Fellowship. More

  • in

    A more effective experimental design for engineering a cell into a new state

    A strategy for cellular reprogramming involves using targeted genetic interventions to engineer a cell into a new state. The technique holds great promise in immunotherapy, for instance, where researchers could reprogram a patient’s T-cells so they are more potent cancer killers. Someday, the approach could also help identify life-saving cancer treatments or regenerative therapies that repair disease-ravaged organs.

    But the human body has about 20,000 genes, and a genetic perturbation could be on a combination of genes or on any of the over 1,000 transcription factors that regulate the genes. Because the search space is vast and genetic experiments are costly, scientists often struggle to find the ideal perturbation for their particular application.   

    Researchers from MIT and Harvard University developed a new, computational approach that can efficiently identify optimal genetic perturbations based on a much smaller number of experiments than traditional methods.

    Their algorithmic technique leverages the cause-and-effect relationship between factors in a complex system, such as genome regulation, to prioritize the best intervention in each round of sequential experiments.

    The researchers conducted a rigorous theoretical analysis to determine that their technique did, indeed, identify optimal interventions. With that theoretical framework in place, they applied the algorithms to real biological data designed to mimic a cellular reprogramming experiment. Their algorithms were the most efficient and effective.

    “Too often, large-scale experiments are designed empirically. A careful causal framework for sequential experimentation may allow identifying optimal interventions with fewer trials, thereby reducing experimental costs,” says co-senior author Caroline Uhler, a professor in the Department of Electrical Engineering and Computer Science (EECS) who is also co-director of the Eric and Wendy Schmidt Center at the Broad Institute of MIT and Harvard, and a researcher at MIT’s Laboratory for Information and Decision Systems (LIDS) and Institute for Data, Systems and Society (IDSS).

    Joining Uhler on the paper, which appears today in Nature Machine Intelligence, are lead author Jiaqi Zhang, a graduate student and Eric and Wendy Schmidt Center Fellow; co-senior author Themistoklis P. Sapsis, professor of mechanical and ocean engineering at MIT and a member of IDSS; and others at Harvard and MIT.

    Active learning

    When scientists try to design an effective intervention for a complex system, like in cellular reprogramming, they often perform experiments sequentially. Such settings are ideally suited for the use of a machine-learning approach called active learning. Data samples are collected and used to learn a model of the system that incorporates the knowledge gathered so far. From this model, an acquisition function is designed — an equation that evaluates all potential interventions and picks the best one to test in the next trial.

    This process is repeated until an optimal intervention is identified (or resources to fund subsequent experiments run out).

    “While there are several generic acquisition functions to sequentially design experiments, these are not effective for problems of such complexity, leading to very slow convergence,” Sapsis explains.

    Acquisition functions typically consider correlation between factors, such as which genes are co-expressed. But focusing only on correlation ignores the regulatory relationships or causal structure of the system. For instance, a genetic intervention can only affect the expression of downstream genes, but a correlation-based approach would not be able to distinguish between genes that are upstream or downstream.

    “You can learn some of this causal knowledge from the data and use that to design an intervention more efficiently,” Zhang explains.

    The MIT and Harvard researchers leveraged this underlying causal structure for their technique. First, they carefully constructed an algorithm so it can only learn models of the system that account for causal relationships.

    Then the researchers designed the acquisition function so it automatically evaluates interventions using information on these causal relationships. They crafted this function so it prioritizes the most informative interventions, meaning those most likely to lead to the optimal intervention in subsequent experiments.

    “By considering causal models instead of correlation-based models, we can already rule out certain interventions. Then, whenever you get new data, you can learn a more accurate causal model and thereby further shrink the space of interventions,” Uhler explains.

    This smaller search space, coupled with the acquisition function’s special focus on the most informative interventions, is what makes their approach so efficient.

    The researchers further improved their acquisition function using a technique known as output weighting, inspired by the study of extreme events in complex systems. This method carefully emphasizes interventions that are likely to be closer to the optimal intervention.

    “Essentially, we view an optimal intervention as an ‘extreme event’ within the space of all possible, suboptimal interventions and use some of the ideas we have developed for these problems,” Sapsis says.    

    Enhanced efficiency

    They tested their algorithms using real biological data in a simulated cellular reprogramming experiment. For this test, they sought a genetic perturbation that would result in a desired shift in average gene expression. Their acquisition functions consistently identified better interventions than baseline methods through every step in the multi-stage experiment.

    “If you cut the experiment off at any stage, ours would still be more efficient than the baselines. This means you could run fewer experiments and get the same or better results,” Zhang says.

    The researchers are currently working with experimentalists to apply their technique toward cellular reprogramming in the lab.

    Their approach could also be applied to problems outside genomics, such as identifying optimal prices for consumer products or enabling optimal feedback control in fluid mechanics applications.

    In the future, they plan to enhance their technique for optimizations beyond those that seek to match a desired mean. In addition, their method assumes that scientists already understand the causal relationships in their system, but future work could explore how to use AI to learn that information, as well.

    This work was funded, in part, by the Office of Naval Research, the MIT-IBM Watson AI Lab, the MIT J-Clinic for Machine Learning and Health, the Eric and Wendy Schmidt Center at the Broad Institute, a Simons Investigator Award, the Air Force Office of Scientific Research, and a National Science Foundation Graduate Fellowship. More

  • in

    Statistics, operations research, and better algorithms

    In this day and age, many companies and institutions are not just data-driven, but data-intensive. Insurers, health providers, government agencies, and social media platforms are all heavily dependent on data-rich models and algorithms to identify the characteristics of the people who use them, and to nudge their behavior in various ways.

    That doesn’t mean organizations are always using optimal models, however. Determining efficient algorithms is a research area of its own — and one where Rahul Mazumder happens to be a leading expert.

    Mazumder, an associate professor in the MIT Sloan School of Management and an affiliate of the Operations Research Center, works both to expand the techniques of model-building and to refine models that apply to particular problems. His work pertains to a wealth of areas, including statistics and operations research, with applications in finance, health care, advertising, online recommendations, and more.

    “There is engineering involved, there is science involved, there is implementation involved, there is theory involved, it’s at the junction of various disciplines,” says Mazumder, who is also affiliated with the Center for Statistics and Data Science and the MIT-IBM Watson AI Lab.

    There is also a considerable amount of practical-minded judgment, logic, and common-sense decision-making at play, in order to bring the right techniques to bear on any individual task.

    “Statistics is about having data coming from a physical system, or computers, or humans, and you want to make sense of the data,” Mazumder says. “And you make sense of it by building models because that gives some pattern to a dataset. But of course, there is a lot of subjectivity in that. So, there is subjectivity in statistics, but also mathematical rigor.”

    Over roughly the last decade, Mazumder, often working with co-authors, has published about 40 peer-reviewed papers, won multiple academic awards, collaborated with major companies about their work, and helped advise graduate students. For his research and teaching, Mazumder was granted tenure by MIT last year.

    From deep roots to new tools

    Mazumder grew up in Kolkata, India, where his father was a professor at the Indian Statistical Institute and his mother was a schoolteacher. Mazumder received his undergraduate and master’s degrees from the Indian Statistical Institute as well, although without really focusing on the same areas as his father, whose work was in fluid mechanics.

    For his doctoral work, Mazumder attended Stanford University, where he earned his PhD in 2012. After a year as a postdoc at MIT’s Operations Research Center, he joined the faculty at Columbia University, then moved to MIT in 2015.

    While Mazumder’s work has many facets, his research portfolio does have notable central achievements. Mazumder has helped combine ideas from two branches of optimization to facilitate addressing computational problems in statistics. One of these branches, discrete optimization, uses discrete variables — integers — to find the best candidate among a finite set of options. This can relate to operational efficiency: What is the shortest route someone might take while making a designated set of stops? Convex optimization, on the other hand, encompasses an array of algorithms that can obtain the best solution for what Mazumder calls “nicely behaved” mathematical functions. They are typically applied to optimize continuous decisions in financial portfolio allocation and health care outcomes, among other things.

    In some recent papers, such as “Fast best subset selection: Coordinate descent and local combinatorial optimization algorithms,” co-authored with Hussein Hazimeh and published in Operations Research in 2020, and in “Sparse regression at scale: branch-and-bound rooted in first-order optimization,” co-authored with Hazimeh and A. Saab and published in Mathematical Programming in 2022, Mazumder has found ways to combine ideas from the two branches.

    “The tools and techniques we are using are new for the class of statistical problems because we are combining different developments in convex optimization and exploring that within discrete optimization,” Mazumder says.

    As new as these tools are, however, Mazumder likes working on techniques that “have old roots,” as he puts it. The two types of optimization methods were considered less separate in the 1950s or 1960s, he says, then grew apart.

    “I like to go back and see how things developed,” Mazumder says. “If I look back in history at [older] papers, it’s actually very fascinating. One thing was developed, another was developed, another was developed kind of independently, and after a while you see connections across them. If I go back, I see some parallels. And that actually helps in my thought process.”

    Predictions and parsimony

    Mazumder’s work is often aimed at simplifying the model or algorithm being applied to a problem. In some instances, bigger models would require enormous amounts of processing power, so simpler methods can provide equally good results while using fewer resources. In other cases — ranging from the finance and tech firms Mazumder has sometimes collaborated with — simpler models may work better by having fewer moving parts.

    “There is a notion of parsimony involved,” Mazumder says. Genomic studies aim to find particularly influential genes; similarly, tech giants may benefit from simpler models of consumer behavior, not more complex ones, when they are recommending a movie to you.

    Very often, Mazumder says, modeling “is a very large-scale prediction problem. But we don’t think all the features or attributes are going to be important. A small collection is going to be important. Why? Because if you think about movies, there are not really 20,000 different movies; there are genres of movies. If you look at individual users, there are hundreds of millions of users, but really they are grouped together into cliques. Can you capture the parsimony in a model?”

    One part of his career that does not lend itself to parsimony, Mazumder feels, is crediting others. In conversation he emphasizes how grateful he is to his mentors in academia, and how much of his work is developed in concert with collaborators and, in particular, his students at MIT. 

    “I really, really like working with my students,” Mazumder says. “I perceive my students as my colleagues. Some of these problems, I thought they could not be solved, but then we just made it work. Of course, no method is perfect. But the fact we can use ideas from different areas in optimization with very deep roots, to address problems of core statistics and machine learning interest, is very exciting.”

    Teaching and doing research at MIT, Mazumder says, allows him to push forward on difficult problems — while also being pushed along by the interest and work of others around him.

    “MIT is a very vibrant community,” Mazumder says. “The thing I find really fascinating is, people here are very driven. They want to make a change in whatever area they are working in. And I also feel motivated to do this.” More

  • in

    Learning the language of molecules to predict their properties

    Discovering new materials and drugs typically involves a manual, trial-and-error process that can take decades and cost millions of dollars. To streamline this process, scientists often use machine learning to predict molecular properties and narrow down the molecules they need to synthesize and test in the lab.

    Researchers from MIT and the MIT-Watson AI Lab have developed a new, unified framework that can simultaneously predict molecular properties and generate new molecules much more efficiently than these popular deep-learning approaches.

    To teach a machine-learning model to predict a molecule’s biological or mechanical properties, researchers must show it millions of labeled molecular structures — a process known as training. Due to the expense of discovering molecules and the challenges of hand-labeling millions of structures, large training datasets are often hard to come by, which limits the effectiveness of machine-learning approaches.

    By contrast, the system created by the MIT researchers can effectively predict molecular properties using only a small amount of data. Their system has an underlying understanding of the rules that dictate how building blocks combine to produce valid molecules. These rules capture the similarities between molecular structures, which helps the system generate new molecules and predict their properties in a data-efficient manner.

    This method outperformed other machine-learning approaches on both small and large datasets, and was able to accurately predict molecular properties and generate viable molecules when given a dataset with fewer than 100 samples.

    “Our goal with this project is to use some data-driven methods to speed up the discovery of new molecules, so you can train a model to do the prediction without all of these cost-heavy experiments,” says lead author Minghao Guo, a computer science and electrical engineering (EECS) graduate student.

    Guo’s co-authors include MIT-IBM Watson AI Lab research staff members Veronika Thost, Payel Das, and Jie Chen; recent MIT graduates Samuel Song ’23 and Adithya Balachandran ’23; and senior author Wojciech Matusik, a professor of electrical engineering and computer science and a member of the MIT-IBM Watson AI Lab, who leads the Computational Design and Fabrication Group within the MIT Computer Science and Artificial Intelligence Laboratory (CSAIL). The research will be presented at the International Conference for Machine Learning.

    Learning the language of molecules

    To achieve the best results with machine-learning models, scientists need training datasets with millions of molecules that have similar properties to those they hope to discover. In reality, these domain-specific datasets are usually very small. So, researchers use models that have been pretrained on large datasets of general molecules, which they apply to a much smaller, targeted dataset. However, because these models haven’t acquired much domain-specific knowledge, they tend to perform poorly.

    The MIT team took a different approach. They created a machine-learning system that automatically learns the “language” of molecules — what is known as a molecular grammar — using only a small, domain-specific dataset. It uses this grammar to construct viable molecules and predict their properties.

    In language theory, one generates words, sentences, or paragraphs based on a set of grammar rules. You can think of a molecular grammar the same way. It is a set of production rules that dictate how to generate molecules or polymers by combining atoms and substructures.

    Just like a language grammar, which can generate a plethora of sentences using the same rules, one molecular grammar can represent a vast number of molecules. Molecules with similar structures use the same grammar production rules, and the system learns to understand these similarities.

    Since structurally similar molecules often have similar properties, the system uses its underlying knowledge of molecular similarity to predict properties of new molecules more efficiently. 

    “Once we have this grammar as a representation for all the different molecules, we can use it to boost the process of property prediction,” Guo says.

    The system learns the production rules for a molecular grammar using reinforcement learning — a trial-and-error process where the model is rewarded for behavior that gets it closer to achieving a goal.

    But because there could be billions of ways to combine atoms and substructures, the process to learn grammar production rules would be too computationally expensive for anything but the tiniest dataset.

    The researchers decoupled the molecular grammar into two parts. The first part, called a metagrammar, is a general, widely applicable grammar they design manually and give the system at the outset. Then it only needs to learn a much smaller, molecule-specific grammar from the domain dataset. This hierarchical approach speeds up the learning process.

    Big results, small datasets

    In experiments, the researchers’ new system simultaneously generated viable molecules and polymers, and predicted their properties more accurately than several popular machine-learning approaches, even when the domain-specific datasets had only a few hundred samples. Some other methods also required a costly pretraining step that the new system avoids.

    The technique was especially effective at predicting physical properties of polymers, such as the glass transition temperature, which is the temperature required for a material to transition from solid to liquid. Obtaining this information manually is often extremely costly because the experiments require extremely high temperatures and pressures.

    To push their approach further, the researchers cut one training set down by more than half — to just 94 samples. Their model still achieved results that were on par with methods trained using the entire dataset.

    “This grammar-based representation is very powerful. And because the grammar itself is a very general representation, it can be deployed to different kinds of graph-form data. We are trying to identify other applications beyond chemistry or material science,” Guo says.

    In the future, they also want to extend their current molecular grammar to include the 3D geometry of molecules and polymers, which is key to understanding the interactions between polymer chains. They are also developing an interface that would show a user the learned grammar production rules and solicit feedback to correct rules that may be wrong, boosting the accuracy of the system.

    This work is funded, in part, by the MIT-IBM Watson AI Lab and its member company, Evonik. More

  • in

    Scaling audio-visual learning without labels

    Researchers from MIT, the MIT-IBM Watson AI Lab, IBM Research, and elsewhere have developed a new technique for analyzing unlabeled audio and visual data that could improve the performance of machine-learning models used in applications like speech recognition and object detection. The work, for the first time, combines two architectures of self-supervised learning, contrastive learning and masked data modeling, in an effort to scale machine-learning tasks like event classification in single- and multimodal data without the need for annotation, thereby replicating how humans understand and perceive our world.

    “A larger portion of human knowledge is learned in a self-supervised way, because we don’t always get supervision signals, and we want to enable the machine-learning model to have the same ability,” says Yuan Gong, an MIT postdoc in the Computer Science and Artificial Intelligence Laboratory (CSAIL).

    “So, another way to put it is that self-supervised learning often forms the foundation of an initial model, because it can learn on vast amounts of unlabeled data. And then you can use classical, supervised learning or reinforcement learning to fine tune the model to something particular if you want to,” says Jim Glass, an MIT senior research scientist and member of the MIT-IBM Watson AI Lab.

    The technique, called the contrastive audio-visual masked autoencoder (CAV-MAE), is a type of neural network that can learn to extract and map meaningful latent representations into high-dimensional space from acoustic and visual data by training on large YouTube datasets of audio and video 10-second clips. The researchers say the technique is more effective than previous approaches because it explicitly models the relationships between audio and visual data in a way that other methods do not.

    Joining Gong and Glass on the study are graduate students Andrew Rouditchenko and Alexander H. Liu of MIT, David Harwath PhD ’18 of the University of Texas at Austin, and MIT-IBM Watson AI Lab members Leonid Karlinsky and Hilde Kuehne. Kuehne is also affiliated with Goethe University Frankfurt. The method was recently presented at the International Conference on Learning Representations.

    A joint and coordinated approach

    The CAV-MAE works by “learning by prediction” and “learning by comparison,” says Gong. The masked data modeling, or the prediction method, takes a video along with its coordinated audio waveform, converts the audio to a spectrogram, and masks 75 percent of both. The unmasked data is tokenized, then fed into separate audio and visual encoders before entering a joint encoder/decoder, where the model is asked to recover the missing data. The difference (reconstruction loss) between the resulting reconstructed prediction and the original audio-visual combination is then used to train the model for better performance. An example of this would be covering part of a video of a piano and part of a spectrogram of piano music, and then asking the model to try to determine the masked inputs. Unfortunately, this method may not capture the association between the video and audio pair, whereas contrastive learning leverages this, but may discard some modality-unique information, like the background in a video.

    Contrastive learning aims to map representations that are similar close to each other. For example, the model will attempt to place different video and audio data of different parrots close to each other and further away from pairs of video and audio of guitars playing. In a similar fashion to masked autoencoding, audio-visual pairs are passed into separate modality encoders; however, the audio and visual components are kept separately within the joint encoder before the model performs pooling and contrastive loss. In this way, contrastive learning tries to identify the parts of each audio or video that are most relevant to the other. For example, if a video shows someone speaking and the corresponding audio clip contains speech, the autoencoder will learn to associate the mouth movements of the speaker with the words being spoken. It will then adjust the model’s parameters so that those inputs are represented close to each other. Ultimately, the CAV-MAE method combines both techniques with multiple forward data streams with masking as a first step, modality-specific encoders, and layer normalization so that the representation strengths are similar.

    “We [then] wanted to compare the proposed CAV-MAE with a model trained only with a masked autoencoder and a model trained only with contrastive learning, because we want to show that by combining masked autoencoder and contrastive learning, we can get some performance improvement,” says Gong, “and the results support our hypothesis that there’s obvious improvement.”

    The researchers tested CAV-MAE — as well as their method without contrastive loss or a masked autoencoder — against other state-of-the-art methods on audio-visual retrieval and audio-visual event classification tasks using standard AudioSet (20K and 2M) and VGGSound datasets — labeled, realistic short clips, which could include multiple sounds. Audio-visual retrieval means that the model sees either the audio or visual component of a query pair and searches for the missing one; event classification includes identifying actions or sounds within data, like a person singing or a car driving.

    Overall, they found that contrastive learning and masked data modeling are complementary methods. CAV-MAE was able to outperform previous techniques (with fully self-supervised pre-training) by about 2 percent for event classification performance verses models with comparable computation and, more impressively, kept pace with or outperformed models with industry-level computational resources. The team’s model ranked similarly to models trained with only the contrastive loss. And surprisingly, the team says, the incorporation of multi-modal data into CAV-MAE pre-training greatly improves the fine-tuning of single-modality representation via supervised learning (with some labeled data) and performance on audio-only event classification tasks. This demonstrates that, like humans, multi-modal information provides an additional “soft label” boost even for audio or visual only tasks; for instance, it helps the model to understand if it’s looking for an electric or acoustic guitar — a richer supervision signal.

    “I think people like the elegance of this model for combining information in the different audio and visual streams. It has the contrastive and the reconstruction loss, and compared to models that have been evaluated with similar data, it clearly does very well across a range of these tasks,” says Glass.

    Building on this, “one special thing is, our model can do both classification and the retrieval, which is not common,” Gong adds. “Before this work, these methods are used separately, but after this work, I see that most of the audio-visual learning frameworks use contracting loss and the masked autoencoder together, implicitly or explicitly.”

    Bringing self-supervised audio-visual learning into our world

    The researchers see their contribution of the contrastive audio-visual masked autoencoder (CAV-MAE) as an important milestone and a step forward for applications, which are increasingly moving from single modality to multi-modality and which require or leverage audio-visual fusion. They hypothesize that one day it could be used for action recognition in realms like sports, education, entertainment, motor vehicles, and public safety. It could also, one day, extend to other modalities. At this time, the fact that, “this only applies to audio-visual data may be a limitation, but we are targeting multi-modal learning, which is trend of machine learning,” says Gong. “As humans, we have multi-modalities — we have smell, touch — many more things that just audio-visual. So, when we try to build AI, we try to mimic humans somehow, not necessarily from the biological perspective, and this method could [potentially be] generalized to other unexplored modalities.”

    As machine-learning models continue to play an increasingly important role in our lives, techniques like this one will become increasingly valuable.

    This research was supported by the MIT-IBM Watson AI Lab. More