multi-task learning
In plain wordsTraining 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.
A training approach in which one model learns several related tasks at once through partly shared internals, so that each task's training signal shapes the shared representations, acting as an inductive biasinductive bias, n.: The assumptions built into a learning method that decide how it fills the gaps between the examples it was shown, such as that similar inputs get similar answers. Every learner needs some to cope with new cases. for the others. With hard parameter sharing, all tasks read from one common stack of layers and only the output heads are task-specific; with soft parameter sharing, each task keeps its own weights and a penalty on their distance holds the task models close. Benefits include better generalizationgeneralization, n.: How well a model copes with new examples it has never seen, rather than just memorizing its training data, like a student who understands the method, not just the answers. and data efficiency, since auxiliary tasks act as a regularizer; lower inference cost than running separate models; and representations that transfer to new tasks. Difficulties include negative transfer, when tasks interfere; balancing losses of different scales and learning speeds, addressed by uncertainty-based weighting, gradient normalization and projection of conflicting gradients; and deciding which tasks to group. Instruction-tuned 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., trained on mixtures of many tasks expressed as text, can be seen as massively multi-task learners, a framing made explicit by text-to-text models such as T5T5, prop. n.: A 2019 Google model that treats every language task — translating, summarizing, answering, classifying — as turning one piece of text into another, using one shared design..
Adding depth estimation as an auxiliary task improved our segmentation model on rare classes.
Origin
Introduced by Rich Caruana in a 1993 conference paper and set out in his 1997 Machine Learning article 'Multitask Learning', which showed that tasks trained in parallel with a shared representation help one another. Ronan Collobert and Jason Weston's unified network for language processing (2008) and Sebastian Ruder's 2017 overview carried the idea into deep learning.
Related
- 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.
- inductive biasinductive bias, n.: The assumptions built into a learning method that decide how it fills the gaps between the examples it was shown, such as that similar inputs get similar answers. Every learner needs some to cope with new cases.
- meta-learningmeta-learning, n.: Training 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.
- instruction tuninginstruction tuning, n.: Training a language model on many tasks written as plain instructions with correct answers, so it learns to do what a request asks instead of just continuing the text.
- catastrophic forgettingcatastrophic forgetting, n.: When a neural network trained on something new gets suddenly and badly worse at what it could already do, because the new training rewrites the internal settings the old skills relied on.
- generalizationgeneralization, n.: How well a model copes with new examples it has never seen, rather than just memorizing its training data, like a student who understands the method, not just the answers.
Sources
- Caruana, R. (1997). Multitask Learning. Machine Learning 28(1).
- Collobert, R. & Weston, J. (2008). A Unified Architecture for Natural Language Processing: Deep Neural Networks with Multitask Learning. Proc. ICML 2008.
- Ruder, S. (2017). An Overview of Multi-Task Learning in Deep Neural Networks. arXiv:1706.05098.
Mentioned in
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., instruction tuninginstruction tuning, n.: Training a language model on many tasks written as plain instructions with correct answers, so it learns to do what a request asks instead of just continuing the text., meta-learningmeta-learning, n.: Training 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., representation learningrepresentation learning, n.: Letting a computer work out for itself how best to describe its data, such as the edges and shapes in images or the meanings of words, instead of relying on descriptions people design by hand., task (machine learning)task (machine learning), n.: The job an AI model is meant to do, such as translating sentences, spotting spam or answering questions, defined by its inputs, its expected outputs and how success is scored., generalist robot policygeneralist robot policy, n.: A single robot-control program trained to do many different tasks in many places, often on several kinds of robots: the opposite of a specialist trained for one job., 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.