没有合适的资源?快使用搜索试试~ 我知道了~
资源详情
资源评论
资源推荐
How Could Machines Learn
Like Animals & Humans?
Harvard MBB
Distinguished Lecture II
2019-03-14
Yann LeCun
Facebook AI Research
New York University
http://yann.lecun.com
Y. LeCun
ML/AI today is mostly supervised learning
Training a machine by showing examples instead of programming it
When the output is wrong, tweak the parameters of the machine
PLANE
CAR
Works well for:
Speech→words
Image→categories
Portrait→ name
Photo→caption
Text→topic
….
Y. LeCun
The Idea goes back to the Perceptron & Adaline
The McCulloch-Pitts Binar Neuron
Perceptron: weights are motorized potentiometers
Adaline: Weights are electrochemical “memistors”
y=sign(
∑
i=1
N
W
i
X
i
+ b)
https://youtu.be/X1G2g3SiCwU
Y. LeCun
The Standard Paradigm of Pattern Recognition
...and “traditional” Machine Learning
Trainable
Classi%er
Feature
Extractor
Hand engineered
Trainable
Y. LeCun
Multilayer Neural Nets and Deep Learning
Traditional Machine Learning
Trainable
Classi%er
Feature
Extractor
Deep Learning
Trainable
Classi%er
Low-Level
Features
Mid-Level
Features
High-Level
Features
Hand engineered
Trainable
Trainable
剩余133页未读,继续阅读
代码深渊漫步者
- 粉丝: 14
- 资源: 320
上传资源 快速赚钱
- 我的内容管理 展开
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功
评论0