def libsvm_train(Kxx, cx, subset, c, pos_weight = 1.0, eps = 1e-3, verbose = 0, probability = 0):
# check input
assert Kxx.shape[0] == Kxx.shape[1] and Kxx.flags.c_contiguous, Kxx.shape
assert subset.flags.c_contiguous and cx.flags.c_contiguous
assert np.all(subset < Kxx.shape[0]) and np.all(subset >= 0)
# set libsvm params
param = libsvm_precomputed.svm_parameter()
libsvm_precomputed.svm_param_set_default(param)
param.nr_weight = 2
param.weight_label = weight_label = yael.ivec(2)
weight_label[0] = -1
weight_label[1] = 1
param.weight = weights = yael.dvec(2)
npos = (cx[subset] == 1).sum()
nneg = (cx[subset] == -1).sum()
weights[0] = 2 * npos / float(npos + nneg)
weights[1] = 2 * nneg / float(npos + nneg) * pos_weight
param.C = c
param.nu = param.p = 0
param.shrinking = 1
param.probability = probability
param.eps = eps
libsvm_precomputed.svm_set_verbose(verbose)
# prepare output
nex = subset.size
dual_coeffs = np.empty((nex,), dtype = np.float64)
bias_out = yael.dvec(3)
# actual call
ret = libsvm_precomputed.svm_train_precomputed(
nex,
yael.numpy_to_ivec_ref(subset),
yael.numpy_to_ivec_ref(cx),
yael.numpy_to_fvec_ref(Kxx),
Kxx.shape[1],
param,
yael.numpy_to_dvec_ref(dual_coeffs),
bias_out)
assert ret > 0
bias_term = bias_out[0]
#print dual_coeffs, bias_term
if probability:
probA = bias_out[1]
probB = bias_out[2]
return dual_coeffs, bias_term, probA, probB
else:
return dual_coeffs, bias_term
elif a == '-d':
d = int(args.pop(0))
elif a == '-nt':
nt = threads.parse_nt(args.pop(0))
else:
print >> sys.stderr, "unknown arg", a
sys.exit(1)
(nb, sizeb, szb) = yutils.vecfile_stats(fb_name, d, fmt_b)
(nq, sizeq, szq) = yutils.vecfile_stats(fq_name, d, fmt_q)
(vb, nb) = yutils.load_vectors_fmt(fb_name, fmt_b, d)
print 'nb=%d db=%d nq=%d ' % (nb, sizeb, nq)
nn = yael.ivec(slice_size * k)
nndis = yael.fvec(slice_size * k)
fnn = open(fnn_name, 'w')
fdis = open(fdis_name, 'w')
t0 = time.time()
for istart in xrange(0, nq, slice_size):
iend = min(istart + slice_size, nq)
nqslice = iend - istart
# Read the queries
(vq, nqslice2) = yutils.load_vectors_fmt(fq_name,
fmt_q,
d,
from yael import yael
yael.common_srandom(12345)
d = 16
n = 1024
k = 4
pts = yael.fvec_new_rand(d * n)
pts = yael.fvec.acquirepointer(pts)
cents = yael.fvec(d * k)
nt = 1
assign = yael.ivec(n)
def d_chi2(a, b):
return (a - b)**2 / (a + b)
print "clustering %d uniform %dD pts in %d centroids" % (n, d, k)
for name, flags in ("L2", 0), ("L1", yael.KMEANS_L1), ("Chi2",
yael.KMEANS_CHI2):
print "%s clustering" % name
t0 = time.time()
flags |= nt | yael.KMEANS_QUIET
from yael import yael
yael.common_srandom(12345)
d=16
n=1024
k=4
pts = yael.fvec_new_rand(d * n)
pts = yael.fvec.acquirepointer(pts)
cents = yael.fvec(d * k)
nt = 1
assign = yael.ivec(n)
def d_chi2(a, b):
return (a - b) ** 2 / (a + b)
print "clustering %d uniform %dD pts in %d centroids" % (n, d, k)
for name, flags in ("L2", 0), ("L1", yael.KMEANS_L1), ("Chi2", yael.KMEANS_CHI2):
print "%s clustering" % name
t0 = time.time()
flags |= nt | yael.KMEANS_QUIET
for run in range(10):
d = int(args.pop(0))
elif a=='-nt':
nt = threads.parse_nt(args.pop(0))
else:
print >> sys.stderr, "unknown arg",a
sys.exit(1)
(nb, sizeb, szb) = yutils.vecfile_stats (fb_name, d, fmt_b)
(nq, sizeq, szq) = yutils.vecfile_stats (fq_name, d, fmt_q)
(vb, nb) = yutils.load_vectors_fmt (fb_name, fmt_b, d)
print 'nb=%d db=%d nq=%d ' % (nb, sizeb, nq)
nn = yael.ivec (slice_size * k)
nndis = yael.fvec (slice_size * k)
fnn = open (fnn_name, 'w')
fdis = open (fdis_name, 'w')
t0 = time.time()
for istart in xrange (0, nq, slice_size):
iend = min (istart + slice_size, nq)
nqslice = iend - istart
# Read the queries
(vq, nqslice2) = yutils.load_vectors_fmt (fq_name, fmt_q, d, nqslice, istart, verbose=False)
assert nqslice == nqslice2