arXivDaily arXiv每日学术速递 周一至周五更新
北美退伍军人节(Veterans Day),12 日无更新
arXiv 2401.07044cs.LG
收藏

BP(λ):通过合成梯度实现在线学习

BP(λ): Online Learning via Synthetic Gradients

University of Bristol(布里斯托大学);University of Oxford(牛津大学)

Joseph Pemberton, Rui Ponte Costa

AI总结:

本文受强化学习 accumulate TD(λ) 启发,提出完全在线的合成梯度学习算法 accumulate BP(λ),通过时序差分误差与资格迹混合控制偏差,避免 BPTT,并在多任务中优于原始方法。

摘要:

训练循环神经网络通常依赖于时间反向传播(BPTT)。BPTT 需要前向和反向传播全部完成,这使得网络在损失梯度可用之前被锁定在这些计算上。最近,Jaderberg 等人提出了合成梯度,以减轻对完整 BPTT 的需求。在他们的实现中,合成梯度通过反向传播梯度和自举合成梯度的混合来学习,类似于强化学习(RL)中的时序差分(TD)算法。然而,与 TD 学习一样,大量使用自举可能会导致偏差,从而导致合成梯度估计不佳。受强化学习中 accumulate TD(λ) 的启发,我们提出了一种完全在线学习合成梯度的方法,完全避免使用 BPTT:accumulate BP(λ)。与 accumulate TD(λ) 一样,我们从分析上证明 accumulate BP(λ) 可以通过使用时序差分误差和递归定义的资格迹的混合来控制偏差水平。接下来,我们通过实验证明,我们的模型在多种任务中学习合成梯度的表现优于原始实现,并且特别适合捕捉更长的时间尺度。最后,基于最近的工作,我们将 accumulate BP(λ) 反思为生物回路中学习的一种原理。总之,受强化学习原理的启发,我们引入了一种能够通过合成梯度实现无偏在线学习的算法。

英文摘要:

Training recurrent neural networks typically relies on backpropagation through time (BPTT). BPTT depends on forward and backward passes to be completed, rendering the network locked to these computations before loss gradients are available. Recently, Jaderberg et al. proposed synthetic gradients to alleviate the need for full BPTT. In their implementation synthetic gradients are learned through a mixture of backpropagated gradients and bootstrapped synthetic gradients, analogous to the temporal difference (TD) algorithm in Reinforcement Learning (RL). However, as in TD learning, heavy use of bootstrapping can result in bias which leads to poor synthetic gradient estimates. Inspired by the accumulate $\mathrm{TD}(λ)$ in RL, we propose a fully online method for learning synthetic gradients which avoids the use of BPTT altogether: accumulate $BP(λ)$. As in accumulate $\mathrm{TD}(λ)$, we show analytically that accumulate $\mathrm{BP}(λ)$ can control the level of bias by using a mixture of temporal difference errors and recursively defined eligibility traces. We next demonstrate empirically that our model outperforms the original implementation for learning synthetic gradients in a variety of tasks, and is particularly suited for capturing longer timescales. Finally, building on recent work we reflect on accumulate $\mathrm{BP}(λ)$ as a principle for learning in biological circuits. In summary, inspired by RL principles we introduce an algorithm capable of bias-free online learning via synthetic gradients.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑