原文：Using human brain activity to guide machine learning
作者： Ruth C. Fong（牛津大學(xué)）肺孵，Walter J. Scheirer（哈佛大學(xué)）
摘要：近年來(lái)贞奋，機(jī)器學(xué)習(xí)在各個(gè)領(lǐng)域獲得了巨大的成功又谋。機(jī)器學(xué)習(xí)向人腦中獲得了很多啟發(fā)，但是卻很少有研究關(guān)于人腦反應(yīng)如何引導(dǎo)機(jī)器學(xué)習(xí)突琳。本文演示了一個(gè)新的‘neurally-weighted’機(jī)器學(xué)習(xí)范例（paradigm）。該范例通過(guò)把參與者fMRI對(duì)每一個(gè)圖片的神經(jīng)反應(yīng)灌輸?shù)皆瓉?lái)的目標(biāo)識(shí)別的訓(xùn)練過(guò)程中，使得學(xué)習(xí)過(guò)程越來(lái)越趨向人腦火焰。
方法
- 從fMRI中獲得激活的權(quán)重
- A. 收集stimulus的激活向量
- 1386彩色圖片。
- ROIs： EBA胧沫，F(xiàn)FA昌简，LO占业，OFA，PPA江场，RSC and TOS
- extrastriate body area (EBA), fusiform face area (FFA), lateral occipital cortex (LO), occiptal face area (OFA), parahippocampal place area (PPA), retrosple- nial cortex (RSC), transverse occipital sulcus (TOS). 1,
- B. 訓(xùn)練分類(lèi)器
- SVM
- C. 獲取激活權(quán)重為到?jīng)Q策線(xiàn)的距離
- 訓(xùn)練圖像分類(lèi)器
- D. 常用的分類(lèi)器訓(xùn)練
- E. 利用激活權(quán)重重新調(diào)整權(quán)重
結(jié)果
通過(guò)將傳統(tǒng)SVM的hinge loss變?yōu)锳ctivity 權(quán)重（AWL）的Loss纺酸，
- C到E分別表示用AWL的EBA，F(xiàn)FA址否，PPA分別改變了某一種或某幾種特定物體的識(shí)別餐蔬。
結(jié)論
Our results provide strong evidence that information measure directly from the human brain can help a machine learning algorithm make better， more human-like decisions and suggest the potential of a new class of hybrid algorithms.
Our paradigm currently requires access to biological data that corresponds to some training examples. Future work should extend this method to other kinds and combinations of biological data, sensory modalities, and machine learning algorithms. Additionally, our current technique discards meaningful information by using scalar activity weight; thus, further research should investigate the development and incorporation of low-dimensional activity weights.
