关于「两部编码账本:一条规则要修正多少条例外才算挣回自己的篇幅」,《规则库的奥卡姆剃刀》给出的定量结论是什么?

结论:第二章建立两部编码账本:一条准确率超过八成、却只净修正 7 条例外的窄规则挣不回自己的篇幅;在合成台账上,门槛从提议者的 12 条规则中只放进 3 条,第三条的余量仅 2.9 个比特。

数据口径:书中的决策记录全部合成、参数均为示意,全部计算由书中程序生成;本书是非个人化的经营分析,不构成经营建议。 第二章建立两部编码账本:一条准确率超过八成、却只净修正 7 条例外的窄规则挣不回自己的篇幅;在合成台账上,门槛从提议者的 12 条规则中只放进 3 条,第三条的余量仅 2.9 个比特。

English version

对照数据表

下表数字由本章 Python 程序原样打印,可重新运行复现。 程序只用 Python 标准库与结论吻合的打印值: 12, 2.9

1347 candidates, N 600, exceptions with no rule 172
kproposed rulenewnetsavecostmargingate
1sg=2&ss=1553757.012.544.5ADMIT
2ch=0&bd=3373164.210.953.3ADMIT
3ss=3&rg=423716.213.22.9ADMIT
4ch=3&rg=41937.211.2-4.0reject
5sg=1&ss=2&rg=45512.313.2-0.9reject
6ch=1&bd=3&rg=4825.113.6-8.6reject
7bd=1&ss=0&rg=4225.115.6-10.5reject
8bd=0&sz=1&rg=1625.213.2-8.0reject
9bd=1&sz=0&rg=4225.313.2-7.9reject
10ss=2&sz=1&rg=1425.413.2-7.8reject
11ch=0&sz=2&rg=2312.713.2-10.5reject
12bd=3&sz=0&rg=4112.713.2-10.5reject
proposer offered 12, gate admitted 3
查看完整验算程序
# Chapter 2. p23: the MDL admission gate on synthetic decision records.
# Records (seeded, illustrative): six attributes ch bd sg ss sz rg with
# V = (4, 4, 3, 4, 3, 5) values; act = 1 with probability 0.85 if
# ch=0 and bd=3, 0.80 if sg=2 and ss=1, 0.45 if rg=4, else 0.12.
# Candidates: every conjunction of 1 to 3 conditions, THEN act.
# A greedy proposer offers, step by step, the candidate that removes the
# most training exceptions; the gate admits while margin > 0 (p21 code).
import itertools
import math
import random

V = (4, 4, 3, 4, 3, 5)
A = len(V)
NAMES = ("ch", "bd", "sg", "ss", "sz", "rg")


def truth(x):
    if x[0] == 0 and x[1] == 3:
        return 0.85
    if x[2] == 2 and x[3] == 1:
        return 0.80
    return 0.45 if x[5] == 4 else 0.12


def records(rng, n):
    out = []
    for _ in range(n):
        x = tuple(rng.randrange(v) for v in V)
        out.append((x, 1 if rng.random() < truth(x) else 0))
    return out


def gamma(n):
    return 2 * (n.bit_length() - 1) + 1


def rule_bits(r):
    m = len(r)
    t = [float(gamma(m)), math.log2(math.comb(A, m))]
    t += [math.log2(V[a]) for a, _ in r]
    return math.fsum(t)


def exc_bits(n, e):
    return gamma(e + 1) + math.log2(math.comb(n, e))


def candidates():
    out = []
    for m in (1, 2, 3):
        for at in itertools.combinations(range(A), m):
            for vs in itertools.product(*[range(V[a]) for a in at]):
                out.append(tuple(zip(at, vs)))
    return out


def cover_masks(data, cands):
    am, pos = {}, 0
    for i, (x, y) in enumerate(data):
        for a in range(A):
            am[(a, x[a])] = am.get((a, x[a]), 0) | (1 << i)
        pos |= y << i
    full = (1 << len(data)) - 1
    cov = []
    for r in cands:
        c = full
        for cond in r:
            c &= am.get(cond, 0)
        cov.append(c)
    return cov, pos, full


def show(r):
    return "&".join("{}={}".format(NAMES[a], v) for a, v in r)


if __name__ == "__main__":
    rng = random.Random(20261113)
    tr = records(rng, 600)
    cands = candidates()
    cov, pos, full = cover_masks(tr, cands)
    n, unc, k = len(tr), full, 0
    e = (pos & full).bit_count()
    print("{} candidates, N {}, exceptions with no rule {}".format(
        len(cands), n, e))
    print("  k  proposed rule      new  net  save   cost  margin  gate")
    admitted = 0
    while True:
        best, bi = 0, -1
        for i, c in enumerate(cov):
            nw = c & unc
            g = (nw & pos).bit_count() - (nw & ~pos).bit_count()
            if g > best:
                best, bi = g, i
        if bi < 0:
            break
        nw = cov[bi] & unc
        save = exc_bits(n, e) - exc_bits(n, e - best)
        cost = rule_bits(cands[bi]) + gamma(k + 2) - gamma(k + 1)
        ok = save > cost and admitted == k
        admitted += 1 if ok else 0
        k, e, unc = k + 1, e - best, unc & ~cov[bi]
        print(" {:>2}  {:<17} {:>4} {:>4} {:>5.1f} {:>6.1f} {:>7.1f}  {}"
              .format(k, show(cands[bi]), nw.bit_count(), best, save,
                      cost, save - cost, "ADMIT" if ok else "reject"))
    print("proposer offered {}, gate admitted {}".format(k, admitted))

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《最小描述长度原则与规则库奥卡姆剃刀、两部编码判定新规则值不值得入库母典》第 2 章「两部编码账本:一条规则要修正多少条例外才算挣回自己的篇幅」(版本 1.0.0). 5D Library. DOI 待发布

DOI 待发布 · 专著页正式发布后上线(本站暂无专著页) · 5DL-B-311 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026311,
  author    = {Fan, Yafu},
  title     = {The Minimum Description Length Principle and Ockham's Razor for Rule Bases: Two-Part Codes that Decide Whether a New Rule Earns Its Place},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论