meta-learning
In plain wordsTraining a system across many different tasks so that it gets better at learning itself, letting it pick up a new task quickly from only a few examples.
Learning to learn: improving a learning algorithm's ability to acquire new tasks by training across a distribution of related tasks rather than on a single one. Training is organized into episodes, each a small task with its own training (support) and evaluation (query) examples, and the meta-learner is optimized for performance after it has adapted to each task. Approaches are commonly grouped as optimization-based, which learn an initialization or update rule that adapts quickly by gradient descent, as in model-agnostic meta-learningmodel-agnostic meta-learning, n.: A training method that searches for a starting point for a model's internal settings from which it can learn any new task of a given kind with just a few small adjustments.; metric-based, which learn an embeddingembedding, n.: A list of numbers that stands for a word, picture or other item, arranged so that similar items get similar lists. It lets a computer measure similarity of meaning as distance. space in which new classes are recognized by comparison with a few labeled examples, as in matching and prototypical networks; and model-based, in which a recurrent or attention-based network reads examples and adapts within its activations. Meta-learning underlies much few-shot learningfew-shot learning, n.: Getting an AI system to handle a new task or recognize a new kind of thing from just a handful of examples, the way a person can learn a new word after hearing it once or twice. research, fast adaptation in reinforcement learningreinforcement learning, n.: A way of teaching a computer program by trial and error: it tries actions, earns points for good results and penalties for bad ones, and gradually learns which choices pay off over time., and AutoMLAutoML, n.: Software that automates the trial-and-error parts of building a machine-learning model, such as choosing the algorithm, its settings and the data-preparation steps, so less expert effort is needed. systems that use results on past datasets to warm-start their searches. The in-context learningin-context learning, n.: A language model's ability to pick up a new task from instructions or examples placed in its input, on the spot, without any retraining — like a quick study learning a game's rules from one round. of large language modelslarge language model, n.: A computer program trained on enormous amounts of text to predict the words that come next. At vast scale, that simple skill lets it chat, answer questions, translate, summarize and write code. is often described as an emergent, model-based form of meta-learning acquired during pretraining.
After meta-learning on 500 related forecasting tasks, the model adapted to a new store from just two weeks of data.
Origin
Jürgen Schmidhuber's 1987 diploma thesis proposed self-referential systems that learn how to learn, and Yoshua Bengio, Samy Bengio and Jocelyn Cloutier (1991) learned synaptic update rules; Sebastian Thrun and Lorien Pratt's 1998 book Learning to Learn collected the early work. The deep-learning revival came in 2016–2017 with memory-augmented networks, matching networks and MAML.
Usage note
Elsewhere 'meta-learning' also names algorithm-selection research that predicts which method suits a dataset, and, in education, learners' awareness of their own learning.
Related
- few-shot learningfew-shot learning, n.: Getting an AI system to handle a new task or recognize a new kind of thing from just a handful of examples, the way a person can learn a new word after hearing it once or twice.
- model-agnostic meta-learningmodel-agnostic meta-learning, n.: A training method that searches for a starting point for a model's internal settings from which it can learn any new task of a given kind with just a few small adjustments.
- in-context learningin-context learning, n.: A language model's ability to pick up a new task from instructions or examples placed in its input, on the spot, without any retraining — like a quick study learning a game's rules from one round.
- transfer learningtransfer learning, n.: Reusing what a model learned on one task to get a head start on another — like a pianist picking up the organ faster than a beginner. It lets models learn new jobs from little data.
- multi-task learningmulti-task learning, n.: Training one model to do several related jobs at once, such as spotting objects and judging distance in the same photo, so that practice on each job can sharpen the others.
- AutoMLAutoML, n.: Software that automates the trial-and-error parts of building a machine-learning model, such as choosing the algorithm, its settings and the data-preparation steps, so less expert effort is needed.
- hypernetworkhypernetwork, n.: A neural network whose job is to produce the settings, or weights, of another neural network, so that one model can generate many specialized models on demand.
Sources
- Schmidhuber, J. (1987). Evolutionary Principles in Self-Referential Learning. Diploma thesis, Technische Universität München.
- Thrun, S. & Pratt, L. (eds.) (1998). Learning to Learn. Boston: Kluwer Academic.
- Hospedales, T., Antoniou, A., Micaelli, P. & Storkey, A. (2022). Meta-Learning in Neural Networks: A Survey. IEEE Transactions on Pattern Analysis and Machine Intelligence 44(9).
Mentioned in
few-shot learningfew-shot learning, n.: Getting an AI system to handle a new task or recognize a new kind of thing from just a handful of examples, the way a person can learn a new word after hearing it once or twice., AutoMLAutoML, n.: Software that automates the trial-and-error parts of building a machine-learning model, such as choosing the algorithm, its settings and the data-preparation steps, so less expert effort is needed., cold start problemcold start problem, n.: The trouble a recommendation system has with brand-new users or items: with no history to learn from, it has little basis for good suggestions., episodeepisode, n.: One complete run of a task from start to finish — a single game of chess, one attempt at a maze — after which everything resets and the agent tries again., in-context learningin-context learning, n.: A language model's ability to pick up a new task from instructions or examples placed in its input, on the spot, without any retraining — like a quick study learning a game's rules from one round., Jürgen SchmidhuberJürgen Schmidhuber, prop. n.: A German computer scientist who co-invented the LSTM, a neural network with memory that powered much of speech recognition and translation in the 2010s, and who argues that early pioneers go under-credited., learninglearning, n.: Getting better at something through experience. For a machine, it means changing its internal settings based on data so that it performs a task better than before., multi-task learningmulti-task learning, n.: Training one model to do several related jobs at once, such as spotting objects and judging distance in the same photo, so that practice on each job can sharpen the others., Chelsea FinnChelsea Finn, prop. n.: An American computer scientist at Stanford known for 'learning to learn' methods that let AI adapt to new tasks from a few examples, and a co-founder of the robotics company Physical Intelligence., hypernetworkhypernetwork, n.: A neural network whose job is to produce the settings, or weights, of another neural network, so that one model can generate many specialized models on demand., mesa-optimizationmesa-optimization, n.: When training an AI produces a model that is itself a goal-seeking planner, with goals of its own that may differ from the ones it was trained for., model-agnosticmodel-agnostic, adj.: Describes a method that works with any kind of AI model, because it only needs to feed the model inputs and observe its outputs, never to look inside., model-agnostic meta-learningmodel-agnostic meta-learning, n.: A training method that searches for a starting point for a model's internal settings from which it can learn any new task of a given kind with just a few small adjustments., nested learningnested learning, n.: A Google research idea that treats an AI model as layers of learning processes updating at different speeds, with the aim of letting models keep learning without forgetting old skills., systematicitysystematicity, n.: The way thinking and language come in connected packages: if you can understand 'the dog chases the cat', you can also understand 'the cat chases the dog'. Whether neural networks manage this has long been debated.