File size: 4,707 Bytes
99f65bd | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 | """
STER — few-shot SCORE FUSION (GPU).
Observation: WM cosine (self-supervised identity, multi-LoD) is strong at K=5
(F1~0.48) but flat; raw-ratio Bagging is weak at K=5 but scales. Fuse them by
rank-averaging so the combined matcher traces the upper envelope across K,
without letting weak features corrupt the tree (no feature concat).
Reports at K in {5,10,20,50}:
bar_raw_bagging baseline
wm_pre_cos WM cosine (frozen pretrain), thr from K labels
fusion_rank 0.5*rank(bagging_prob) + 0.5*rank(wm_cos), thr from K labels
fusion_max elementwise max of the two rank scores
"""
import os, json, numpy as np, torch, torch.nn as nn, torch.nn.functional as F
from sklearn.ensemble import BaggingClassifier
from sklearn.preprocessing import StandardScaler
from sklearn.metrics import f1_score
from scipy.stats import rankdata
NPZ = "/root/ster/data/ster_wm_vectors.npz"
OUT = "/root/ster/exp/ster_fusion_results.json"
DEV = "cuda" if torch.cuda.is_available() else "cpu"
torch.manual_seed(1); np.random.seed(1)
z = np.load(NPZ, allow_pickle=True)
lod12, lod22, candX, indexX = z["lod12_X"], z["lod22_X"], z["cand_X"], z["index_X"]
trP, trY, teP, teY = z["train_pairs"], z["train_y"], z["test_pairs"], z["test_y"]
DIM = lod12.shape[1]
scaler = StandardScaler().fit(np.vstack([lod12, lod22, candX, indexX]))
def T(x): return torch.tensor(scaler.transform(x), dtype=torch.float32, device=DEV)
L12, L22, CAND, INDEX = T(lod12), T(lod22), T(candX), T(indexX)
class Enc(nn.Module):
def __init__(s, d):
super().__init__(); s.net = nn.Sequential(nn.Linear(d,64),nn.GELU(),nn.Linear(64,64),nn.GELU(),nn.Linear(64,32))
def forward(s,x): return F.normalize(s.net(x),dim=-1)
def info_nce(a,b,tau=0.1):
lg=a@b.t()/tau; lab=torch.arange(a.size(0),device=a.device)
return 0.5*(F.cross_entropy(lg,lab)+F.cross_entropy(lg.t(),lab))
enc=Enc(DIM).to(DEV); opt=torch.optim.Adam(enc.parameters(),1e-3,weight_decay=1e-5)
for ep in range(400):
p=torch.randperm(L12.size(0),device=DEV)
for i in range(0,L12.size(0),256):
b=p[i:i+256]
if b.numel()<8: continue
loss=info_nce(enc(L12[b]),enc(L22[b])); opt.zero_grad(); loss.backward(); opt.step()
enc.eval()
with torch.no_grad():
ec=enc(CAND).cpu().numpy(); ei=enc(INDEX).cpu().numpy()
def ratio(P):
a,b=candX[P[:,0]],indexX[P[:,1]]
with np.errstate(divide='ignore',invalid='ignore'):
r=np.where(b!=0,a/b,1000.0)
return np.clip(np.round(r,3),None,1000.0)
raw_tr,raw_te=ratio(trP),ratio(teP)
cos_tr=np.sum(ec[trP[:,0]]*ei[trP[:,1]],1); cos_te=np.sum(ec[teP[:,0]]*ei[teP[:,1]],1)
def thr_from(scores, y):
return float(np.median([np.quantile(scores[y==1],0.2) if (y==1).any() else .5,
np.quantile(scores[y==0],0.8) if (y==0).any() else .5]))
Ks,NDRAW=[5,10,20,50],15
pos=np.where(trY==1)[0]; neg=np.where(trY==0)[0]
report={}
for K in Ks:
acc={k:[] for k in ["bar_raw_bagging","wm_pre_cos","fusion_rank","fusion_max"]}
for rep in range(NDRAW):
r=np.random.RandomState(300+rep)
pidx=r.choice(pos,K,replace=False); nidx=r.choice(neg,min(2*K,len(neg)),replace=False)
idx=np.concatenate([pidx,nidx]); y=trY[idx]
clf=BaggingClassifier(n_estimators=50,random_state=1).fit(raw_tr[idx],y)
prob_te=clf.predict_proba(raw_te)[:,1]; prob_tr=clf.predict_proba(raw_tr[idx])[:,1]
acc["bar_raw_bagging"].append(f1_score(teY,(prob_te>=0.5).astype(int),zero_division=0))
tcos=thr_from(cos_tr[idx],y)
acc["wm_pre_cos"].append(f1_score(teY,(cos_te>=tcos).astype(int),zero_division=0))
# rank-fusion on test set
rb=rankdata(prob_te)/len(prob_te); rc=rankdata(cos_te)/len(cos_te)
fr=0.5*rb+0.5*rc; fm=np.maximum(rb,rc)
# thresholds for fused scores from the K training pairs
rb_tr=rankdata(prob_tr)/len(prob_tr)
# map train cos to test-rank scale via quantile of cos among test
rc_tr=np.array([ (cos_te<cv).mean() for cv in cos_tr[idx] ])
fr_tr=0.5*rb_tr+0.5*rc_tr; fm_tr=np.maximum(rb_tr,rc_tr)
acc["fusion_rank"].append(f1_score(teY,(fr>=thr_from(fr_tr,y)).astype(int),zero_division=0))
acc["fusion_max"].append(f1_score(teY,(fm>=thr_from(fm_tr,y)).astype(int),zero_division=0))
report[K]={k:dict(f1=round(float(np.mean(v)),4),std=round(float(np.std(v)),4)) for k,v in acc.items()}
print(f"\n=== K={K} ===",flush=True)
for k,s in sorted(report[K].items(),key=lambda kv:-kv[1]['f1']):
print(f" {k:20s} F1={s['f1']:.4f} ±{s['std']:.4f}",flush=True)
os.makedirs(os.path.dirname(OUT),exist_ok=True); json.dump(report,open(OUT,"w"),indent=2)
print("\nSaved ->",OUT,flush=True)
|