arXiv research

Neural networks learn symbolic structure to perform compositional tasks.

OCEAN infers online task identities from context variables.

A generally intelligent learner should generalize to more complex tasks than it has previously encountered, but the two common paradigms in machine learning -- either training a separate learner per task or training a single learner for all tasks -- both have difficulty with such generalization because they do not leve…

NeSS combines neural and symbolic approaches for better compositional generalization.

Transformers can generalize to a large task family with only a few demonstrations.

Study on when RLVR can learn compositional problems.

This paper explores optimising acquisition functions in Bayesian optimisation.

CPVs enable efficient learning and generalization of complex tasks.

SCL discovers compositional structures in analogical reasoning tasks.

New algorithm reduces complexity for optimizing complex machine learning tasks.

Framework for lifelong learning of compositional structures.

Study challenges neural models in compositional learning tasks.

Simple model outperforms neural networks on language understanding tasks.

Unified analysis of multi-task functional linear regression with manifold and composite penalties.

A Boolean algebra formalizes task composition for reinforcement learning.

Transformers learn to solve modular arithmetic tasks by in-context learning and skill composition.

Agents compose pre-trained policies for complex tasks, improving zero-shot performance.

The successful application of general reinforcement learning algorithms to real-world robotics applications is often limited by their high data requirements. We introduce Regularized Hierarchical Policy Optimization (RHPO) to improve data-efficiency for domains with multiple dominant tasks and ultimately reduce require…

Lower bounds set for infinite-precision transformers.

Since its inception, the modus operandi of multi-task learning (MTL) has been to minimize the task-wise mean of the empirical risks. We introduce a generalized loss-compositional paradigm for MTL that includes a spectrum of formulations as a subfamily. One endpoint of this spectrum is minimax MTL: a new MTL formulation…

This paper analyzes how diffusion models learn and generalize concepts.

Solving complex, temporally-extended tasks is a long-standing problem in reinforcement learning (RL). We hypothesize that one critical element of solving such problems is the notion of compositionality. With the ability to learn concepts and sub-skills that can be composed to solve longer tasks, i.e. hierarchical RL, w…

Proposes a framework for compositional generalization in language models.

Transformer architecture struggles with complex tasks due to limitations in function composition.

RTE enables extrapolation to new tasks by learning task transformations.

RICH models scenes as hierarchical tree to learn and generate complex compositions.

UCB-TQL learns from multiple tasks with shared dynamics and adapts to task-specific variations.

This work proves composite neural networks outperform their components under certain conditions.

New composite indicators reveal hidden relationships between indicators.

Describes explaining neurons in deep representations using compositional logical concepts.

Unified theory for neural scaling laws in hierarchically compositional data.

HydaLearn dynamically adjusts task weights for better MTL performance.

New samplers improve compositional generation with diffusion models.

An important property for lifelong-learning agents is the ability to combine existing skills to solve unseen tasks. In general, however, it is unclear how to compose skills in a principled way. We provide a "recipe" for optimal value function composition in entropy-regularised reinforcement learning (RL) and then exten…

Paper analyzes stability and generalization of SCO algorithms.

Embedding models for entities and relations are extremely useful for recovering missing facts in a knowledge base. Intuitively, a relation can be modeled by a matrix mapping entity vectors. However, relations reside on low dimension sub-manifolds in the parameter space of arbitrary matrices---for one reason, compositio…

In this work, we propose a composition/decomposition framework for adversarially training generative models on composed data - data where each sample can be thought of as being constructed from a fixed number of components. In our framework, samples are generated by sampling components from component generators and fee…

We consider evidence integration from potentially dependent observation processes under varying spatio-temporal sampling resolutions and noise levels. We develop a multi-resolution multi-task (MRGP) framework while allowing for both inter-task and intra-task multi-resolution and multi-fidelity. We develop shallow Gauss…

Study shows how transformers learn to combine simple tasks into complex ones.

Transformers learn to generalize unseen tasks by composing self-attention layers.

The composition of elementary behaviors to solve challenging transfer learning problems is one of the key elements in building intelligent machines. To date, there has been plenty of work on learning task-specific policies or skills but almost no focus on composing necessary, task-agnostic skills to find a solution to …

Humans are able to perform a myriad of sophisticated tasks by drawing upon skills acquired through prior experience. For autonomous agents to have this capability, they must be able to extract reusable skills from past experience that can be recombined in new ways for subsequent tasks. Furthermore, when controlling com…

Standard methods in deep learning for natural language processing fail to capture the compositional structure of human language that allows for systematic generalization outside of the training distribution. However, human learners readily generalize in this way, e.g. by applying known grammatical rules to novel words.…

This paper improves neural network explanations by quantifying and visualizing semantic compositions.

This paper evaluates how well neural models can solve complex tasks by breaking them into simpler ones.

We introduce Compositional Imitation Learning and Execution (CompILE): a framework for learning reusable, variable-length segments of hierarchically-structured behavior from demonstration data. CompILE uses a novel unsupervised, fully-differentiable sequence segmentation module to learn latent encodings of sequential d…

Compositionality is a key strategy for addressing combinatorial complexity and the curse of dimensionality. Recent work has shown that compositional solutions can be learned and offer substantial gains across a variety of domains, including multi-task learning, language modeling, visual question answering, machine comp…

Modular neural networks generalize better with less data.