Biologically plausible local synaptic learning rules robustly implement deep supervised learning.
Biologically plausible local synaptic learning rules robustly implement deep supervised learning.
复制标题
DOI：
10.3389/fnins.2023.1160899
复制
发表时间：
2023
影响因子：
4.3
通讯作者：
Miura, Keiji
中科院分区：
文献类型：
--
作者：
Konishi, Masataka;Igarashi, Kei M.;Miura, Keiji
关键词：
在深度神经网络中，中间层的表征学习对于实现高效学习至关重要。然而，当前流行的反向传播学习规则（BP）在生物学上不一定合理，并且无法以其当前形式在大脑中实施。因此，为了阐明大脑使用的学习规则，为实际记忆任务建立生物学上合理的学习规则至关重要。例如，导致学习性能比实验研究中观察到的动物差的学习规则可能不是在真实大脑中使用的计算，应该被排除。通过数值模拟，我们开发了生物学上合理的学习规则来解决重复实验室实验的任务，在该实验中，小鼠学会了预测正确的奖励金额。尽管极限学习机（ELM）和权重扰动（WP）学习规则的表现比小鼠差，但反馈对齐（FA）规则的表现与BP相当。为了获得生物学上更合理的模型，我们开发了 FA 的变体 FA_Ex-100%，它实现直接多巴胺输入，在焦点层（如小鼠内嗅皮层中发现的）局部提供误差信号。 FA_Ex-100%的性能与传统BP相当。最后，我们测试了 FA_Ex-100% 对于规则扰动和生物学上不可避免的噪声是否具有鲁棒性。即使受到扰动，FA_Ex-100% 也能发挥作用，大概是因为如果扰动产生偏差，它可以在下一步中校准正确的预测误差（例如多巴胺能信号）作为教学信号。这些结果表明，当可能由多巴胺能神经元传递的误差信号准确时，简化且生物学上合理的学习规则（例如 FA_Ex-100%）可以有力地促进深度监督学习。三种学习规则示意图：BP、反向传播； FA，反馈对齐； FA_Ex-100%，与中间层100%兴奋性神经元的反馈对齐。 BP需要W2中的信息来进行反向传播。 FA 要求中间层神经元对输出的初步影响具有异质性。 FA_Ex-100% 在生物学上是最合理的，因为它只能使用本地可用的信息在突触三联体上进行计算，如下所述，但其性能相当好，可与 BP 相媲美。使用符号 、 和 ，BP 的梯度由 给出，其中 FA 被随机数 Bi 替代，FA_Ex100% 被 1 替代。因此，对于 FA_Ex-100%，中间层的突触权重按以下规则更新：其中 θ 是阶跃函数，Ii 是神经元 i 的当前输入。这可以解释为 (ΔW)ij ∝ prei × postj × 多巴胺。有趣的是，只要误差信号（可能由多巴胺能神经元传递）准确，像 FA_Ex-100% 这样的简化且生物学上合理的学习规则就能稳健地工作。
In deep neural networks, representational learning in the middle layer is essential for achieving efficient learning. However, the currently prevailing backpropagation learning rules (BP) are not necessarily biologically plausible and cannot be implemented in the brain in their current form. Therefore, to elucidate the learning rules used by the brain, it is critical to establish biologically plausible learning rules for practical memory tasks. For example, learning rules that result in a learning performance worse than that of animals observed in experimental studies may not be computations used in real brains and should be ruled out. Using numerical simulations, we developed biologically plausible learning rules to solve a task that replicates a laboratory experiment where mice learned to predict the correct reward amount. Although the extreme learning machine (ELM) and weight perturbation (WP) learning rules performed worse than the mice, the feedback alignment (FA) rule achieved a performance equal to that of BP. To obtain a more biologically plausible model, we developed a variant of FA, FA_Ex-100%, which implements direct dopamine inputs that provide error signals locally in the layer of focus, as found in the mouse entorhinal cortex. The performance of FA_Ex-100% was comparable to that of conventional BP. Finally, we tested whether FA_Ex-100% was robust against rule perturbations and biologically inevitable noise. FA_Ex-100% worked even when subjected to perturbations, presumably because it could calibrate the correct prediction error (e.g., dopaminergic signals) in the next step as a teaching signal if the perturbation created a deviation. These results suggest that simplified and biologically plausible learning rules, such as FA_Ex-100%, can robustly facilitate deep supervised learning when the error signal, possibly conveyed by dopaminergic neurons, is accurate. Schematic illustration of three learning rules: BP, backpropagation; FA, feedback alignment; and FA_Ex-100%, feedback alignment with 100% excitatory neurons in middle layer. BP requires the information in W2 to backprop. FA requires heterogeneity in the tentative impact of the middle layer neurons on the output. FA_Ex-100% is the most biologically plausible in the sense that it can be computed at a synaptic triad only with locally available information as explained below, but its performance is fairly good and comparable to that of BP. With the notations, , , and , the gradient for BP is given by whose is replaced by a random number Bi for FA and by 1 for FA_Ex100%. Therefore, for FA_Ex-100%, the synaptic weights in the middle layer are updated by the following rule: where θ is a step function and Ii is the current input to neuron i. This can be interpreted as (ΔW)ij ∝ prei × postj × dopamine. Interestingly, simplified and biologically plausible learning rules like FA_Ex-100% work robustly as far as the error signal, possibly conveyed by dopaminergic neurons, is accurate.
登录
查看更多内容
DOI：
10.1073/pnas.2100600119
发表时间：
2022-03-15
影响因子：
11.1
作者：
Hiratani N;Latham PE
通讯作者：
Latham PE
影响因子：
2.9
作者：
Nakazono T;Jun H;Blurton-Jones M;Green KN;Igarashi KM
通讯作者：
Igarashi KM
影响因子：
3
作者：
Nakazono T;Lam TN;Patel AY;Kitazawa M;Saito T;Saido TC;Igarashi KM
通讯作者：
Igarashi KM
影响因子：
25
作者：
Haddad, Rafi;Lanjuin, Anne;Madisen, Linda;Zeng, Hongkui;Murthy, Venkatesh N.;Uchida, Naoshige
通讯作者：
Uchida, Naoshige
影响因子：
6
作者：
Huang, Guang-Bin;Zhu, Qin-Yu;Siew, Chee-Kheong
通讯作者：
Siew, Chee-Kheong