Learning computational tasks from single examples

New “meta-learning” approach improves on the state of the art in “one-shot” learning.

In the past decade, deep-learning systems have proven remarkably successful at many artificial-intelligence tasks, but their applications tend to be narrow. A computer vision system trained to recognize cats and dogs, for instance, would need significant retraining to start recognizing sharks and sea turtles.

Meta-learning is a paradigm intended to turn machine learning systems into generalists. A meta-learning model is trained on a range of related tasks, but it learns not only how to perform those tasks but also how to learn to perform them. The idea is that it could then be adapted to new tasks with only a handful of labeled training examples, drastically reducing the need for labor-intensive data annotation.

At the (virtual) International Conference on Learning Representations, we will present an approach that improves performance on meta-learning tasks without increasing the data annotation requirements. The key idea is to adapt the meta-learning procedure so that it can leverage small sets of unlabeled data, in addition to the traditional labeled examples.

Meta-learning
In meta-learning, a machine learning model learns how to learn. During meta-training, the model is trained on a group of related tasks — using data from “support sets” — and tested using data from “query sets”. But the query sets are labeled, so the model can assess how effectively it's learning. During meta-testing, the model is again trained on a group of support sets, but it's evaluated on its ability to classify unlabeled query data.
Stacy Reilly

The intuition is that even without labels, these extra data still contain a lot of useful information. Suppose, for instance, that a meta-learning system trained on images of terrestrial animals (such as cats and dogs) is being adapted to recognize aquatic animals. Unlabeled images of aquatic animals (i.e., images that don’t indicate whether an animal is a shark or a sea turtle) still tell the model something about the learning task, such as the lighting conditions and background colors typical of underwater photos.

In experiments, we compared models trained through our approach to 16 different baselines on an object recognition meta-learning task. We found that our approach improved performance on one-shot learning, or learning a new object classification task from only a single labeled example, by 11% to 16%, depending on the architectures of the underlying neural networks.

Meta-learning

In conventional machine learning, a model is fed a body of labeled data and learns to correlate data features with the labels. Then it’s fed a separate body of test data and evaluated on how well it predicts the labels for that data. For evaluation purposes, the system designers have access to the test-data labels, but the model itself doesn’t.

Meta-learning adds another layer of complexity. During meta-training — the analogue of conventional training — the model learns to perform a range of related tasks. Each task has its own sets of training data and test data, and the model sees both. That is, part of its meta-training is learning how particular ways of responding to training data tend to affect its performance on test data.

During meta-testing, it is again trained on a range of tasks. These are related to but not identical to the tasks it saw during meta-training — recognizing aquatic animals, for instance, as opposed to terrestrial animals. Again, for each task, the model sees both training data and test data. But whereas, during meta-training, the test data were labeled, during meta-testing, the labels are unknown and must be predicted.

The terminology can get a bit confusing, so meta-learning researchers typically refer to the meta-learning “training” sets as support sets and the meta-learning “test” sets as query sets. During meta-training, the learning algorithm has access to the labels for both the support sets and the query sets, and it uses them to produce a global model. During meta-testing, it has access only to the labels for the support sets, which it uses to adapt the global model to each of the new tasks.

Our approach has two key innovations. First, during meta-training, we do not learn a single global model. Instead, we train an auxiliary neural network to produce a local model for each task, based on the corresponding support set. Second and more important, during meta-training we also train a second auxiliary network to leverage the unlabeled data of the query sets. Then, during meta-testing, we can use the query sets to fine-tune the local models, improving performance.

Leveraging unlabeled data

A machine learning system is governed by a set of parameters, and in meta-learning, meta-training optimizes them for a particular family of tasks — such as recognizing animals. During meta-testing or operational deployment, the model uses a handful of training examples to optimize those parameters for a new task.

A particular set of parameter values defines a point in a multidimensional space, and adaptation to a new task can be thought of as searching the space for the point representing the optimal new settings.

Meta-learning parameter space
In traditional meta-learning (left), the result of training is a model (φ) that can be adapted to a new set of related tasks (1 – 4). Adaptation involves searching for the optimal settings 1 – θ4) of the model parameters, based on a small set of labeled data (dl1 – dl4). Our system (right), by contrast, uses the labeled data and the available unlabeled data (x1 – x4) to better approximate those settings.

A traditional meta-learning system might begin its search at the point defined by the global model (φ in the figure above); this is the initialization step. Then, using the labeled data of the support set, it would work its way toward the settings that correspond to the new task; this is the adaptation step.

With our approach, by contrast, the initialization network selects a starting search location on the basis of the data in the support set 01(dl1) – θ04(dl4) in the figure above). Then it works its way toward the optimal settings using the unlabeled data of the query set (x1 – x4, above). More precisely, the second auxiliary neural network estimates the gradient implied by the query set data.

In the same way that the parameter settings of a machine learning model can be interpreted as a point in a representational space, so can the parameter settings and the resulting error rate on a particular data set. The multidimensional graph that results is like a topological map, with depressions that represent low error rates and peaks that represent high error rates. In this context, machine learning is a matter of identifying the slope of a depression — a gradient — and moving down it, toward a region of low error.

This is how many machine learning systems learn, but typically, they have the advantage of knowing, from training data labels, what the true error rate is for a given set of system parameters. In our case, because we’re relying on the unlabeled query set data, we can only guess at the true gradients.

That’s where the second auxiliary neural network comes in: it infers gradients from query set data. The system as a whole then uses the inferred gradients to fine-tune the initial parameter settings supplied by the first neural network.

The approach can be explained and justified through connections to two topics in theoretical machine learning, namely empirical Bayes and information bottleneck. These theoretical developments are beyond the scope of this blog post, but the interested reader can consult the full manuscript.

The associated software code has also been open-sourced as part of the Xfer repository.

Although our system beat all 16 baselines on the task of one-shot learning, there were several baseline systems that outperformed it on five-shot learning, or learning with five examples per new task. The approaches used by those baselines are complementary to our approach, and we believe that combining approaches could yield even lower error rates. Going forward, that’s one of several extensions of this work that we will be pursuing.

Research areas

Related content

US, WA, Seattle
Note that this posting is for a handful of teams within Amazon Robotics. Teams include: Robotics, Computer Vision, Machine Learning, Optimization, and more.Are you excited about building high-performance robotic systems that can perceive and learn to help deliver for customers? The Amazon Robotics team is creating new science products and technologies that make this possible, at Amazon scale. We work at the intersection of computer vision, machine learning, robotic manipulation, navigation, and human-robot interaction.Amazon Robotics is seeking broad, curious applied scientists and engineering interns to join our diverse, full-stack team. In addition to designing, building, and delivering end-to-end robotic systems, our team is responsible for core infrastructure and tools that serve as the backbone of our robotic applications, enabling roboticists, applied scientists, software and hardware engineers to collaborate and deploy systems in the lab and in the field. We will give you the tools and support you need to invent with us in ways that are rewarding, fulfilling and fun. Come join us!A day in the lifeAs an intern you will develop a new algorithm to solve one of the challenging computer vision and manipulation problems in Amazon's robotic warehouses. Your project will fit your academic research experience and interests. You will code and test out your solutions in increasingly realistic scenarios and iterate on the idea with your mentor to find the best solution to the problem.
US, WA, Bellevue
The Global Supply Chain-ACES organization aims to raise the bar on Amazon’s customer experience by delivering holistic solutions for Global Customer Fulfillment that facilitate the effective and efficient movement of product through our supply chain. We develop strategies, processes, material handling and technology solutions, reporting and other mechanisms, which are simple, technology enabled, globally scalable, and locally relevant. We achieve this through cross-functional partnerships, listening to the needs of our customers and prioritizing initiatives to deliver maximum impact across the value chain. Within the organization, our Quality team balances tactical operation with operations partners with global engagement on programs to deliver improved inventory accuracy in our network. The organization is looking for an experienced Principal Data Scientist to partner with senior leadership to develop long term strategic solutions. As a Principal Scientist, they will lead critical initiatives for Global Supply Chain, leveraging complex data analysis and visualization to:a. Collaborate with business teams to define data requirements and processes;b. Automate data pipelines;c. Design, develop, and maintain scalable (automated) reports and dashboards that track progress towards plans;d. Define, track and report program success metrics.e. Serve as a technical science lead on our most demanding, cross-functional projects.
US, MA, North Reading
Are you inspired by invention? Is problem solving through teamwork in your DNA? Do you like the idea of seeing how your work impacts the bigger picture? Answer yes to any of these and you’ll fit right in here at Amazon Robotics. We are a smart team of doers that work passionately to apply cutting edge advances in robotics and software to solve real-world challenges that will transform our customers’ experiences in ways we can’t even imagine yet. We invent new improvements every day. We are Amazon Robotics and we will give you the tools and support you need to invent with us in ways that are rewarding, fulfilling and fun. The Research Science team at Amazon Robotics is seeking interns with a passion for robotic research to work on cutting edge algorithms for robotics. Our team works on challenging and high-impact projects, including allocating resources to complete a million orders a day, coordinating the motion of thousands of robots, autonomous navigation in warehouses, and learning how to grasp all the products Amazon sells. We are seeking internship candidates with backgrounds in computer vision, machine learning, resource allocation, discrete optimization, search, planning/scheduling, and reinforcement learning. As an intern you will develop a new algorithm to solve one of the challenging computer vision and manipulation problems in Amazon's robotic warehouses. Your project will fit your academic research experience and interests. You will code and test out your solutions in increasingly realistic scenarios and iterate on the idea with your mentor to find the best solution to the problem.
US, WA, Seattle
Are you excited about building high-performance robotic systems that can perceive, learn, and act intelligently alongside humans? The Robotics AI team is creating new science products and technologies that make this possible, at Amazon scale. We work at the intersection of computer vision, machine learning, robotic manipulation, navigation, and human-robot interaction.The Amazon Robotics team is seeking broad, curious applied scientists and engineering interns to join our diverse, full-stack team. In addition to designing, building, and delivering end-to-end robotic systems, our team is responsible for core infrastructure and tools that serve as the backbone of our robotic applications, enabling roboticists, applied scientists, software and hardware engineers to collaborate and deploy systems in the lab and in the field. Come join us!
US, WA, Bellevue
Employer: Amazon.com Services LLCPosition: Research Scientist IILocation: Bellevue, WA Multiple Positions Available1. Research, build and implement highly effective and innovative methods in Statistical Modeling, Machine Learning, and other quantitative techniques such as operational research and optimization to deliver algorithms that solve real business problems.2. Take initiative to scope and plan research projects based on roadmap of business owners and enable data-driven solutions. Participate in shaping roadmap for the research team.3. Ensure data quality throughout all stages of acquisition and processing of the data, including such areas as data sourcing/collection, ground truth generation, data analysis, experiment, evaluation and visualization etc.4. Navigate a variety of data sources, understand the business reality behind large-scale data and develop meaningful science solutions.5. Partner closely with product or/and program owners, as well as scientists and engineers in cross-functional teams with a clear path to business impact and deliver on demanding projects.6. Present proposals and results in a clear manner backed by data and coupled with conclusions to business customers and leadership team with various levels of technical knowledge, educating them about underlying systems, as well as sharing insights.7. Perform experiments to validate the feature additions as requested by domain expert teams.8. Some telecommuting benefits available.The pay range for this position in Bellevue, WA is $136,000-$184,000 (yr); however, base pay offered may vary depending on job-related knowledge, skills, and experience. A sign-on bonus and restricted stock units may be provided as part of the compensation package, in addition to a full range of medical, financial, and/or other benefits, dependent on the position offered. This information is provided by the Washington Equal Pay Act. Base pay information is based on market location. Applicants should apply via Amazon's internal or external careers site.#0000
US, VA, Arlington
The Central Science Team within Amazon’s People Experience and Technology org (PXTCS) uses economics, behavioral science, statistics, and machine learning to proactively identify mechanisms and process improvements which simultaneously improve Amazon and the lives, well-being, and the value of work to Amazonians. We are an interdisciplinary team, which combines the talents of science and engineering to develop and deliver solutions that measurably achieve this goal. As Director for PXT Central Science Technology, you will be responsible for leading multiple teams through rapidly evolving complex demands and define, develop, deliver and execute on our science roadmap and vision. You will provide thought leadership to scientists and engineers to invent and implement scalable machine learning recommendations and data driven algorithms supporting flexible UI frameworks. You will manage and be responsible for delivering some of our most strategic technical initiatives. You will design, develop and operate new, highly scalable software systems that support Amazon’s efforts to be Earth’s Best Employer and have a significant impact on Amazon’s commitment to our employees and communities where we both serve and employ 1.3 million Amazonians. As Director of Applied Science, you will be part of the larger technical leadership community at Amazon. This community forms the backbone of the company, plays a critical role in the broad business planning, works closely with senior executives to develop business targets and resource requirements, influences our long-term technical and business strategy, helps hire and develop engineering leaders and developers, and ultimately enables us to deliver engineering innovations.This role is posted for Arlington, VA, but we are flexible on location at many of our offices in the US and Canada.
US, VA, Arlington
Employer: Amazon.com Services LLCPosition: Data Scientist IILocation: Arlington, VAMultiple Positions Available1. Manage and execute entire projects or components of large projects from start to finish including data gathering and manipulation, synthesis and modeling, problem solving, and communication of insights and recommendations.2. Oversee the development and implementation of data integration and analytic strategies to support population health initiatives.3. Leverage big data to explore and introduce areas of analytics and technologies.4. Analyze data to identify opportunities to impact populations.5. Perform advanced integrated comprehensive reporting, consultative, and analytical expertise to provide healthcare cost and utilization data and translate findings into actionable information for internal and external stakeholders.6. Oversee the collection of data, ensuring timelines are met, data is accurate and within established format.7. Act as a data and technical resource and escalation point for data issues, ensuring they are brought to resolution.8. Serve as the subject matter expert on health care benefits data modeling, system architecture, data governance, and business intelligence tools. #0000
US, TX, Dallas
Employer: Amazon.com Services LLCPosition: Data Scientist II (multiple positions available)Location: Dallas, TX Multiple Positions Available:1. Assist customers to deliver Machine Learning (ML) and Deep Learning (DL) projects from beginning to end, by aggregating data, exploring data, building and validating predictive models, and deploying completed models to deliver business impact to the organization;2. Apply understanding of the customer’s business need and guide them to a solution using AWS AI Services, AWS AI Platforms, AWS AI Frameworks, and AWS AI EC2 Instances;3. Use Deep Learning frameworks like MXNet, PyTorch, Caffe 2, Tensorflow, Theano, CNTK, and Keras to help our customers build DL models;4. Research, design, implement and evaluate novel computer vision algorithms and ML/DL algorithms;5. Work with data architects and engineers to analyze, extract, normalize, and label relevant data;6. Work with DevOps engineers to help customers operationalize models after they are built;7. Assist customers with identifying model drift and retraining models;8. Research and implement novel ML and DL approaches, including using FPGA;9. Develop computer vision and machine learning methods and algorithms to address real-world customer use-cases; and10. Design and run experiments, research new algorithms, and work closely with engineers to put algorithms and models into practice to help solve customers' most challenging problems.11. Approximately 15% domestic and international travel required.12. Telecommuting benefits are available.#0000
US, WA, Seattle
MULTIPLE POSITIONS AVAILABLECompany: AMAZON.COM SERVICES LLCPosition Title: Manager III, Data ScienceLocation: Bellevue, WashingtonPosition Responsibilities:Manage a team of data scientists working to build large-scale, technical solutions to increase effectiveness of Amazon Fulfillment systems. Define key business goals and map them to the success of technical solutions. Aggregate, analyze and model data from multiple sources to inform business decisions. Manage and quantify improvement in the customer experience resulting from research outcomes. Develop and manage a long-term research vision and portfolio of research initiatives, with algorithms and models that to be integrated in production systems. Hire and mentor junior scientists.Amazon.com is an Equal Opportunity-Affirmative Action Employer – Minority / Female / Disability / Veteran / Gender Identity / Sexual Orientation #0000
US, VA, Arlington
MULTIPLE POSITIONS AVAILABLECompany: AMAZON.COM SERVICES LLCPosition Title: Data Scientist IILocation: Arlington, VirginiaPosition Responsibilities:Design and implement scalable and reliable approaches to support or automate decision making throughout the business. Apply a range of data science techniques and tools combined with subject matter expertise to solve difficult business problems and cases in which the solution approach is unclear. Acquire data by building the necessary SQL / ETL queries. Import processes through various company specific interfaces for accessing Oracle, RedShift, and Spark storage systems. Build relationships with stakeholders and counterparts. Analyze data for trends and input validity by inspecting univariate distributions, exploring bivariate relationships, constructing appropriate transformations, and tracking down the source and meaning of anomalies. Build models using statistical modeling, mathematical modeling, econometric modeling, network modeling, social network modeling, natural language processing, machine learning algorithms, genetic algorithms, and neural networks. Validate models against alternative approaches, expected and observed outcome, and other business defined key performance indicators. Implement models that comply with evaluations of the computational demands, accuracy, and reliability of the relevant ETL processes at various stages of production.Amazon.com is an Equal Opportunity-Affirmative Action Employer – Minority / Female / Disability / Veteran / Gender Identity / Sexual Orientation #0000