> For the complete documentation index, see [llms.txt](https://www.bfm-unity.org/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://www.bfm-unity.org/bfm-urs-wei-mian-jian-meng-lian-he-ti-chu-bei-xi-tong-qu-kuai-lian-jia-mi-huo-bi-hong-guan-tiao-kong/the-tzolkin-calendar/zhuo-er-jin-li-fa-de-sheng-ji-1-ni-he-han-shu.md).

# 卓尔金升级1——拟合函数

{% hint style="success" %}
**智能版 3.0 —— 智能函数拟合，更精确，更易调节**

[卓尔金-智能原子钟](https://share.weiyun.com/y9kahvvr) | [卓尔金-智能飞船](https://share.weiyun.com/NrrPpvCw)

紧跟时代步伐，引领行业变革。
{% endhint %}

## 智能神经网络查找表

```python
=$E$4 + ($AI$4-$E$4) * 1/(1+EXP(-($D$4*(F3-$C$2))))
 
=$D$5 - ($D$5-$D$35) * 1/(1+EXP(-($D$4*(-C6+$B$3))))

=(E$4+$D5)/2

=MAX((F$4+$D5)/2,E4)

=MIN((E$4+$D6)/2,D5)

=IFS(
    ($C6-$B$3)<-(F$3-$C$2), MIN((F$4+$D6)/2,E5),
    ($C6-$B$3)>-(F$3-$C$2), MAX((F$4+$D6)/2,E5),
    TRUE, (F$4+$D6)/2
)
```

```c
激活函数

sigmoid(x) = 1/(1 + exp(−x))

tanh(x) = (exp(x) - exp(−x))/(exp(x) + exp(−x))
tanh(x) = (1 - exp(−2 * x))/(1 + exp(−2 * x))

tanh(x) = 2 * sigmoid(2 * x) - 1
tanh(x / 2) = 2 * sigmoid(x) - 1 

ReLU(x) = MAX(x,0)

Swish(x) = x * sigmoid(x)
```