-1

Due to some reason the following code only predicts 1 for any input after training on a dataset. Please help. Have already tried lower learning rate, training in batches(even one at a time), not using initial weights as zero,using reduce_mean,etc.

xtf = tf.placeholder(tf.float32, [None,7])
w = tf.Variable(tf.truncated_normal([7,1], stddev=1./math.sqrt(7)))
b = tf.Variable(tf.zeros([1]))
ytf = tf.nn.softmax(tf.matmul(xtf, w) + b)
y_ = tf.placeholder(tf.float32, [None,1])
cross_entropy = tf.reduce_mean(-tf.reduce_sum(y_ * tf.log(ytf),reduction_indices =[1]))
train_step = tf.train.GradientDescentOptimizer(0.001).minimize(cross_entropy)
sess = tf.InteractiveSession()
tf.global_variables_initializer().run()
ptr=0
batch = 1
x,y,xt = data()
while (ptr+100) <=(len(x)-1):
    x_batch, y_batch = x[ptr:ptr+100], y[ptr:ptr+100]
    sess.run(train_step, feed_dict={xtf: x_batch, y_: y_batch})
    ptr += 100
predict = sess.run(ytf, feed_dict={xtf: xt})

xtf = [[0.2711736617240513, 0.014151057562208049, 0.0, 0.125, 1.0, 1.0, 0.0], [0.4722292033174164, 0.13913573538264068, 0.0, 0.125, 0.0, 0.0, 1.0], [0.32143754712239253, 0.015468569817999833, 0.0, 0.0, 1.0, 1.0, 1.0], [0.43453128926866047, 0.10364429745562033, 0.0, 0.125, 1.0, 0.0, 1.0], [0.43453128926866047, 0.015712553569072387, 0.0, 0.0, 1.0, 1.0, 0.0], [0.3465694898215632, 0.01650950209357577, 0.0, 0.0, 0.5, 1.0, 0.0], [0.6732847449107816, 0.10122885832000206, 0.0, 0.0, 1.0, 0.0, 0.0], [0.01985423473234481, 0.04113566043083236, 0.16666666666666666, 0.375, 1.0, 1.0, 0.0], [0.33400351847197784, 0.021730754366528396, 0.3333333333333333, 0.0, 1.0, 1.0, 1.0], [0.17064589092736868, 0.058694292654020104, 0.0, 0.125, 0.0, 0.5, 1.0], [0.04498617743151546, 0.03259622914329302, 0.16666666666666666, 0.125, 1.0, 1.0, 1.0], [0.7235486303091229, 0.051822148727810165, 0.0, 0.0, 1.0, 0.0, 1.0], [0.24604171902488062, 0.015712553569072387, 0.0, 0.0, 1.0, 1.0, 0.0], [0.4847951746670017, 0.06104473451835265, 0.8333333333333334, 0.125, 1.0, 1.0, 0.0], [0.17064589092736868, 0.015330377421392339, 0.0, 0.0, 1.0, 1.0, 1.0], [0.6858507162603669, 0.03122992013728673, 0.0, 0.0, 1.0, 0.5, 1.0], [0.01985423473234481, 0.056848213999904744, 0.16666666666666666, 0.5, 0.5, 1.0, 0.0], [0.3465694898215632, 0.025374310111545468, 0.0, 0.0, 1.0, 0.5, 0.0], [0.38426740387031916, 0.03513366015444757, 0.0, 0.125, 1.0, 1.0, 1.0], [0.3465694898215632, 0.014102260811993537, 0.0, 0.0, 0.0, 1.0, 1.0], [0.43453128926866047, 0.050748620223090936, 0.0, 0.0, 1.0, 0.5, 0.0], [0.42196531791907516, 0.025374310111545468, 0.0, 0.0, 1.0, 0.5, 0.0], [0.18321186227695402, 0.015671954672893913, 0.0, 0.0, 0.5, 1.0, 1.0], [0.3465694898215632, 0.06929138530460492, 0.0, 0.0, 1.0, 0.0, 0.0], [0.09525006282985675, 0.04113566043083236, 0.16666666666666666, 0.375, 1.0, 1.0, 1.0], [0.4722292033174164, 0.061264319894317944, 0.8333333333333334, 0.125, 1.0, 1.0, 1.0], [0.3465694898215632, 0.014102260811993537, 0.0, 0.0, 0.0, 1.0, 0.0], [0.23347574767529528, 0.5133418122566505, 0.3333333333333333, 0.375, 1.0, 0.0, 0.0], [0.3465694898215632, 0.01537917417160685, 0.0, 0.0, 0.5, 1.0, 1.0], [0.3465694898215632, 0.015411575213749284, 0.0, 0.0, 1.0, 1.0, 0.0], [0.4973611460165871, 0.05410739813385612, 0.0, 0.0, 0.0, 0.0, 0.0], [0.3465694898215632, 0.2859895551532101, 0.0, 0.125, 0.0, 0.0, 1.0], [0.3465694898215632, 0.015126992566498259, 0.0, 0.0, 0.5, 1.0, 1.0], [0.8240764011058055, 0.020494635090094415, 0.0, 0.0, 1.0, 0.5, 0.0], [0.3465694898215632, 0.16038672010106, 0.0, 0.125, 0.0, 0.0, 0.0], [0.5224930887157577, 0.10149724044618187, 0.0, 0.125, 1.0, 0.0, 0.0], [0.3465694898215632, 0.014110458666029575, 0.0, 0.0, 0.0, 1.0, 0.0], [0.2586076903744659, 0.015712553569072387, 0.0, 0.0, 1.0, 1.0, 0.0], [0.22090977632570996, 0.03513366015444757, 0.0, 0.25, 1.0, 1.0, 1.0], [0.17064589092736868, 0.021942337075458514, 0.0, 0.125, 0.0, 1.0, 1.0], [0.4973611460165871, 0.018493968331299484, 0.0, 0.125, 1.0, 1.0, 1.0], [0.33400351847197784, 0.04098927018018883, 0.0, 0.125, 1.0, 0.5, 1.0], [0.3465694898215632, 0.015411575213749284, 0.0, 0.0, 0.0, 1.0, 0.0], [0.032420206081930136, 0.08115719346076702, 0.3333333333333333, 0.125, 0.0, 0.5, 1.0], [0.23347574767529528, 0.01537917417160685, 0.0, 0.0, 0.5, 1.0, 1.0], [0.3465694898215632, 0.015712553569072387, 0.0, 0.0, 1.0, 1.0, 0.0], [0.3465694898215632, 0.030253985132996517, 0.0, 0.125, 0.5, 1.0, 0.0], [0.3465694898215632, 0.015126992566498259, 0.0, 0.0, 0.5, 1.0, 1.0], [0.3465694898215632, 0.04231498029001666, 0.0, 0.25, 0.0, 1.0, 0.0], [0.22090977632570996, 0.034743286152731485, 0.0, 0.125, 1.0, 1.0, 1.0], [0.08268409148027142, 0.07746484096553544, 0.16666666666666666, 0.5, 1.0, 1.0, 0.0], [0.2586076903744659, 0.01522458606692728, 0.0, 0.0, 1.0, 1.0, 0.0], [0.610454888162855, 0.1497654242623688, 0.0, 0.125, 0.0, 0.0, 1.0], [0.35913546117114853, 0.050748620223090936, 0.0, 0.125, 1.0, 0.5, 1.0], [0.8115104297562201, 0.1209753416358076, 0.16666666666666666, 0.0, 0.0, 0.0, 0.0], [0.3465694898215632, 0.06929138530460492, 0.0, 0.0, 1.0, 0.0, 0.0], [0.2586076903744659, 0.020494635090094415, 0.0, 0.0, 1.0, 0.5, 1.0], [0.35285247549635584, 0.014110458666029575, 0.0, 0.0, 0.0, 1.0, 0.0], [0.05755214878110078, 0.05416439273810667, 0.3333333333333333, 0.125, 1.0, 0.5, 1.0], [0.13294797687861273, 0.09154270340242172, 0.3333333333333333, 0.625, 1.0, 1.0, 0.0], [0.2711736617240513, 0.014110458666029575, 0.0, 0.0, 0.0, 1.0, 0.0], [0.4722292033174164, 0.15614960068643363, 0.0, 0.0, 1.0, 0.0, 1.0], [0.5601910027645137, 0.16293234896625058, 0.0, 0.125, 1.0, 0.0, 0.0], [0.04498617743151546, 0.05445717323939373, 0.3333333333333333, 0.375, 1.0, 1.0, 0.0], [0.3465694898215632, 0.05410739813385612, 0.0, 0.0, 0.0, 0.0, 0.0], [0.3465694898215632, 0.029757819776815374, 0.16666666666666666, 0.125, 0.0, 1.0, 0.0], [0.35913546117114853, 0.020494635090094415, 0.0, 0.0, 1.0, 0.5, 1.0], [0.23347574767529528, 0.015923941091001644, 0.0, 0.0, 1.0, 1.0, 0.0], [0.20834380497612465, 0.015468569817999833, 0.3333333333333333, 0.5, 1.0, 1.0, 1.0], [0.32143754712239253, 0.016908073949327893, 0.0, 0.25, 1.0, 1.0, 0.0], [0.39683337521990447, 0.020494635090094415, 0.0, 0.0, 1.0, 0.5, 0.0], [0.19577783362653933, 0.09154270340242172, 0.3333333333333333, 0.625, 1.0, 1.0, 1.0], [0.2586076903744659, 0.1434624456306609, 0.0, 0.0, 1.0, 0.5, 0.0], [0.32143754712239253, 0.028212719478023115, 0.0, 0.125, 0.0, 1.0, 0.0], [0.39683337521990447, 0.11027245763075773, 0.0, 0.0, 1.0, 1.0, 0.0], [0.3088715757728072, 0.014931805565640218, 0.0, 0.0, 1.0, 1.0, 0.0], [0.3465694898215632, 0.015411575213749284, 0.0, 0.0, 1.0, 1.0, 0.0], [0.3465694898215632, 0.015712553569072387, 0.0, 0.0, 1.0, 1.0, 0.0], [0.005152048253329982, 0.056604230248832196, 0.3333333333333333, 0.0, 1.0, 0.5, 0.0], [0.37170143252073384, 0.024349578357040744, 0.0, 0.0, 1.0, 1.0, 1.0], [0.2711736617240513, 0.017566830077223785, 0.0, 0.0, 1.0, 1.0, 0.0], [0.35913546117114853, 0.018542765081513996, 0.0, 0.0, 1.0, 1.0, 0.0], [0.3465694898215632, 0.015200187691820024, 0.0, 0.0, 0.5, 1.0, 1.0], [0.3465694898215632, 0.09193307740413781, 0.0, 0.0, 1.0, 0.0, 0.0], [0.20834380497612465, 0.020494635090094415, 0.0, 0.0, 1.0, 0.5, 1.0], [0.4093993465694898, 0.030937139635999665, 0.0, 0.375, 1.0, 1.0, 1.0], [0.19577783362653933, 0.06709553154495196, 0.5, 0.125, 1.0, 1.0, 0.0], [0.3465694898215632, 0.015712553569072387, 0.0, 0.0, 1.0, 1.0, 0.0], [0.2837396330736366, 0.5133418122566505, 0.3333333333333333, 0.375, 1.0, 0.0, 1.0], [0.2963056044232219, 0.015712553569072387, 0.0, 0.0, 1.0, 1.0, 0.0], [0.35913546117114853, 0.015712553569072387, 0.0, 0.0, 1.0, 1.0, 0.0], [0.24604171902488062, 0.015330377421392339, 0.0, 0.0, 1.0, 1.0, 0.0], [0.572756974114099, 0.11940564777490723, 0.0, 0.125, 1.0, 0.0, 0.0], [0.32143754712239253, 0.04015972542654215, 0.3333333333333333, 0.125, 1.0, 1.0, 0.0], [0.7361146016587082, 0.014151057562208049, 0.0, 0.0, 1.0, 1.0, 0.0], [0.3465694898215632, 0.015712553569072387, 0.0, 0.0, 1.0, 1.0, 0.0], [0.8869062578537321, 0.06764049365134761, 0.0, 0.0, 0.0, 0.0, 0.0], [0.2837396330736366, 0.12366716556464086, 0.16666666666666666, 0.0, 0.0, 0.0, 0.0], [0.42196531791907516, 0.04489301019734967, 0.16666666666666666, 0.0, 1.0, 0.5, 1.0], [0.42196531791907516, 0.050748620223090936, 0.0, 0.125, 1.0, 0.5, 0.0], [0.3465694898215632, 0.015411575213749284, 0.0, 0.0, 1.0, 1.0, 1.0], [0.3465694898215632, 0.015411575213749284, 0.0, 0.0, 1.0, 1.0, 0.0], [0.2586076903744659, 0.15085515328815924, 0.16666666666666666, 0.0, 1.0, 0.0, 0.0], [0.4093993465694898, 0.016891873428256675, 0.0, 0.0, 1.0, 1.0, 0.0], [0.4596632319678311, 0.015468569817999833, 0.0, 0.25, 1.0, 1.0, 0.0], [0.3465694898215632, 0.015411575213749284, 0.0, 0.0, 1.0, 1.0, 0.0], [0.2586076903744659, 0.014931805565640218, 0.0, 0.0, 1.0, 1.0, 1.0], [0.3465694898215632, 0.015175789316712771, 0.0, 0.0, 1.0, 1.0, 0.0], [0.4722292033174164, 0.015411575213749284, 0.0, 0.0, 1.0, 1.0, 0.0], [0.3465694898215632, 0.04713766070721715, 0.0, 0.125, 0.5, 1.0, 1.0], [0.5853229454636844, 0.10149724044618187, 0.0, 0.0, 1.0, 0.0, 0.0], [0.17692887660216136, 0.028212719478023115, 0.0, 0.125, 0.0, 1.0, 1.0], [0.2711736617240513, 0.015712553569072387, 0.0, 0.0, 1.0, 1.0, 0.0], [0.24604171902488062, 0.019177122834302632, 0.0, 0.125, 1.0, 1.0, 1.0], [0.20834380497612465, 0.028220722145058292, 0.0, 0.0, 0.0, 1.0, 1.0], [0.2586076903744659, 0.015468569817999833, 0.0, 0.0, 1.0, 1.0, 0.0], [0.8806232721789394, 0.015126992566498259, 0.0, 0.0, 0.5, 1.0, 0.0], [0.35913546117114853, 0.04098927018018883, 0.0, 0.125, 1.0, 0.5, 0.0], [0.2963056044232219, 0.4831284260198326, 0.16666666666666666, 0.0, 0.0, 0.0, 0.0], [0.01985423473234481, 0.06104473451835265, 0.3333333333333333, 0.5, 1.0, 1.0, 1.0], [0.2586076903744659, 0.1434624456306609, 0.0, 0.25, 1.0, 0.5, 0.0], [0.3465694898215632, 0.015712553569072387, 0.0, 0.0, 1.0, 1.0, 0.0], [0.40311636089469716, 0.058694292654020104, 0.0, 0.125, 0.0, 0.5, 0.0], [0.40311636089469716, 0.025374310111545468, 0.0, 0.0, 1.0, 0.5, 1.0], [0.6732847449107816, 0.15085515328815924, 0.16666666666666666, 0.0, 1.0, 0.0, 0.0], [0.14551394822819805, 0.021942337075458514, 0.0, 0.125, 0.0, 1.0, 0.0], [0.3465694898215632, 0.015126992566498259, 0.0, 0.0, 0.5, 1.0, 0.0], [0.2963056044232219, 0.01393967004027879, 0.0, 0.0, 1.0, 1.0, 0.0], [0.3465694898215632, 0.04364049521284361, 0.16666666666666666, 0.125, 0.0, 1.0, 1.0], [0.5601910027645137, 0.013614293309848433, 0.0, 0.0, 1.0, 1.0, 0.0], [0.4093993465694898, 0.015411575213749284, 0.0, 0.0, 0.0, 1.0, 0.0], [0.24604171902488062, 0.013760683560491965, 0.0, 0.0, 1.0, 1.0, 0.0], [0.5853229454636844, 0.028302115124416098, 0.0, 0.125, 1.0, 1.0, 1.0], [0.35913546117114853, 0.050748620223090936, 0.0, 0.125, 1.0, 0.5, 1.0], [0.3088715757728072, 0.025374310111545468, 0.0, 0.0, 1.0, 0.5, 0.0], [0.2837396330736366, 0.02936744577509929, 0.0, 0.0, 0.0, 0.5, 0.0], [0.23347574767529528, 0.051301584996521765, 0.3333333333333333, 0.0, 1.0, 0.0, 1.0], [0.4596632319678311, 0.10364429745562033, 0.0, 0.125, 1.0, 0.0, 0.0], [0.19577783362653933, 0.01798980030808316, 0.0, 0.0, 1.0, 1.0, 0.0], [0.2963056044232219, 0.15458810467956932, 0.0, 0.0, 0.0, 0.0, 0.0], [0.3465694898215632, 0.029757819776815374, 0.3333333333333333, 0.0, 0.0, 1.0, 1.0], [0.2711736617240513, 0.015126992566498259, 0.0, 0.0, 1.0, 1.0, 1.0], [0.2963056044232219, 0.030937139635999665, 0.0, 0.125, 1.0, 1.0, 1.0], [0.23347574767529528, 0.013175122557917838, 0.0, 0.0, 0.5, 1.0, 0.0], [0.22090977632570996, 0.022446505098674834, 0.0, 0.0, 1.0, 0.5, 0.0], [0.23347574767529528, 0.07173122281533045, 0.16666666666666666, 0.125, 1.0, 0.5, 0.0], [0.33400351847197784, 0.015216388212891242, 0.0, 0.0, 1.0, 1.0, 0.0], [0.10781603417944208, 0.06709553154495196, 0.3333333333333333, 0.25, 1.0, 1.0, 1.0], [0.4533802462930384, 0.050748620223090936, 0.3333333333333333, 0.0, 1.0, 0.5, 0.0], [0.5224930887157577, 0.025374310111545468, 0.0, 0.0, 1.0, 0.5, 0.0], [0.6355868308620256, 0.02444717185746977, 0.0, 0.0, 1.0, 0.5, 0.0], [0.2711736617240513, 0.12999454257145598, 0.0, 0.125, 1.0, 0.0, 1.0], [0.6921337019351596, 0.015712553569072387, 0.0, 0.0, 1.0, 1.0, 0.0], [0.5036441316913798, 0.028302115124416098, 0.3333333333333333, 0.0, 1.0, 1.0, 0.0], [0.3465694898215632, 0.014273049437744325, 0.0, 0.0, 1.0, 1.0, 0.0], [0.6355868308620256, 0.11980421963065935, 0.16666666666666666, 0.0, 0.0, 0.0, 0.0], [0.19577783362653933, 0.015094396337354966, 0.0, 0.0, 0.5, 1.0, 1.0], [0.37170143252073384, 0.015712553569072387, 0.0, 0.0, 1.0, 1.0, 0.0], [0.3465694898215632, 0.016908073949327893, 0.0, 0.0, 1.0, 1.0, 0.0], [0.3465694898215632, 0.13575255909676823, 0.3333333333333333, 1.0, 1.0, 1.0, 0.0], [0.5476250314149284, 0.031425107138144774, 0.16666666666666666, 0.0, 1.0, 1.0, 0.0], [0.4973611460165871, 0.030741952635141623, 0.0, 0.0, 1.0, 0.5, 1.0], [0.32143754712239253, 0.015175789316712771, 0.0, 0.0, 1.0, 1.0, 0.0], [0.20834380497612465, 0.016908073949327893, 0.0, 0.0, 1.0, 1.0, 0.0], [0.007288263382759489, 0.07746484096553544, 0.16666666666666666, 0.5, 1.0, 1.0, 0.0], [0.10781603417944208, 0.04006213192611313, 0.3333333333333333, 0.0, 1.0, 1.0, 0.0], [0.3465694898215632, 0.10735285047192313, 0.16666666666666666, 0.0, 1.0, 0.0, 1.0], [0.5601910027645137, 0.05445717323939373, 0.6666666666666666, 0.125, 1.0, 1.0, 1.0], [0.3465694898215632, 0.050602229972447406, 0.0, 0.0, 1.0, 0.0, 0.0], [0.3465694898215632, 0.11027245763075773, 0.0, 0.0, 1.0, 1.0, 0.0], [0.7612465443578789, 0.06538764528744409, 0.0, 0.0, 1.0, 0.0, 0.0], [0.04498617743151546, 0.056848213999904744, 0.16666666666666666, 0.5, 0.5, 1.0, 0.0], [0.007288263382759489, 0.021730754366528396, 0.16666666666666666, 0.125, 1.0, 1.0, 1.0], [0.2586076903744659, 0.015468569817999833, 0.0, 0.0, 1.0, 1.0, 0.0], [0.6984166876099522, 0.05991421140938287, 0.0, 0.0, 0.0, 0.0, 0.0], [0.22090977632570996, 0.015330377421392339, 0.16666666666666666, 0.125, 1.0, 1.0, 0.0], [0.3465694898215632, 0.04970768794751499, 0.16666666666666666, 0.375, 1.0, 1.0, 0.0], [0.6230208595124404, 0.05604306762136532, 0.0, 0.0, 0.0, 0.0, 1.0], [0.37170143252073384, 0.025374310111545468, 0.0, 0.0, 1.0, 0.5, 0.0], [0.4470972606182458, 0.0, 0.0, 0.0, 1.0, 1.0, 0.0], [0.3465694898215632, 0.13575255909676823, 0.3333333333333333, 1.0, 1.0, 1.0, 1.0], [0.3465694898215632, 0.02937564362913533, 0.0, 0.0, 0.0, 0.5, 0.0], [0.10781603417944208, 0.061264319894317944, 0.3333333333333333, 0.5, 1.0, 1.0, 0.0], [0.007288263382759489, 0.0761229303346364, 0.16666666666666666, 0.25, 1.0, 0.5, 0.0], [0.04498617743151546, 0.04298993693898376, 0.3333333333333333, 0.0, 1.0, 1.0, 1.0], [0.3465694898215632, 0.09759350042902103, 0.0, 0.0, 1.0, 0.0, 0.0], [0.3465694898215632, 0.030253985132996517, 0.0, 0.125, 0.5, 1.0, 1.0], [0.5601910027645137, 0.051822148727810165, 0.0, 0.0, 1.0, 0.0, 0.0], [0.4973611460165871, 0.030253985132996517, 0.16666666666666666, 0.125, 0.5, 1.0, 0.0], [0.4470972606182458, 0.015411575213749284, 0.0, 0.0, 1.0, 1.0, 0.0], [0.39683337521990447, 0.025374310111545468, 0.0, 0.0, 1.0, 0.5, 1.0], [0.23347574767529528, 0.025374310111545468, 0.0, 0.0, 1.0, 0.5, 0.0], [0.23347574767529528, 0.015330377421392339, 0.0, 0.125, 1.0, 1.0, 1.0], [0.032420206081930136, 0.050748620223090936, 0.16666666666666666, 0.125, 1.0, 0.5, 0.0], [0.5476250314149284, 0.05410739813385612, 0.0, 0.0, 0.0, 0.0, 1.0], [0.7235486303091229, 0.2859895551532101, 0.0, 0.0, 0.0, 0.0, 1.0], [0.3465694898215632, 0.015126992566498259, 0.0, 0.0, 0.5, 1.0, 0.0], [0.5224930887157577, 0.01640390592611157, 0.16666666666666666, 0.0, 1.0, 1.0, 0.0], [0.3465694898215632, 0.015126992566498259, 0.0, 0.0, 0.5, 1.0, 1.0], [0.2963056044232219, 0.025374310111545468, 0.0, 0.0, 1.0, 0.5, 1.0]]

ytf = [[0.0], [1.0], [1.0], [1.0], [0.0], [0.0], [0.0], [0.0], [1.0], [1.0], [1.0], [1.0], [0.0], [0.0], [0.0], [1.0], [0.0], [1.0], [0.0], [1.0], [0.0], [1.0], [1.0], [1.0], [0.0], [1.0], [0.0], [0.0], [1.0], [0.0], [0.0], [1.0], [1.0], [0.0], [0.0], [0.0], [1.0], [0.0], [0.0], [1.0], [0.0], [0.0], [0.0], [1.0], [1.0], [0.0], [0.0], [1.0], [0.0], [0.0], [0.0], [0.0], [1.0], [1.0], [0.0], [1.0], [1.0], [0.0], [1.0], [0.0], [0.0], [1.0], [0.0], [0.0], [0.0], [1.0], [1.0], [0.0], [1.0], [0.0], [0.0], [0.0], [0.0], [0.0], [1.0], [0.0], [0.0], [0.0], [1.0], [1.0], [0.0], [1.0], [1.0], [0.0], [1.0], [1.0], [0.0], [0.0], [1.0], [0.0], [0.0], [0.0], [0.0], [0.0], [0.0], [0.0], [0.0], [1.0], [1.0], [0.0], [0.0], [0.0], [0.0], [0.0], [0.0], [0.0], [1.0], [1.0], [0.0], [1.0], [0.0], [0.0], [0.0], [0.0], [0.0], [0.0], [0.0], [0.0], [0.0], [0.0], [0.0], [0.0], [0.0], [1.0], [0.0], [1.0], [0.0], [1.0], [1.0], [0.0], [0.0], [0.0], [0.0], [1.0], [0.0], [0.0], [1.0], [0.0], [0.0], [0.0], [0.0], [1.0], [1.0], [0.0], [0.0], [0.0], [1.0], [0.0], [0.0], [0.0], [0.0], [1.0], [0.0], [0.0], [0.0], [0.0], [1.0], [0.0], [0.0], [0.0], [0.0], [1.0], [0.0], [0.0], [0.0], [1.0], [1.0], [0.0], [0.0], [0.0], [0.0], [0.0], [1.0], [0.0], [0.0], [0.0], [0.0], [0.0], [0.0], [0.0], [0.0], [0.0], [0.0], [1.0], [1.0], [0.0], [1.0], [1.0], [0.0], [0.0], [1.0], [0.0], [1.0], [1.0], [1.0], [1.0], [0.0], [0.0], [1.0], [0.0]]

xt(test data) = [[0.45272319662402744, 0.015281580671177828, 0.0, 0.0, 0.5, 1.0, 0.0], [0.6175656072794409, 0.013663090060062943, 0.0, 0.125, 1.0, 1.0, 1.0], [0.815376500065937, 0.018908740708122825, 0.0, 0.0, 0.5, 0.5, 0.0], [0.3538177502307794, 0.016908073949327893, 0.0, 0.0, 1.0, 1.0, 0.0], [0.287880785968614, 0.023983602730431916, 0.1111111111111111, 0.125, 1.0, 1.0, 1.0], [0.18238164314914942, 0.01800600082915438, 0.0, 0.0, 1.0, 1.0, 0.0], [0.3933799287880786, 0.014891206669461744, 0.0, 0.0, 0.5, 1.0, 1.0], [0.3406303573783463, 0.056604230248832196, 0.1111111111111111, 0.125, 1.0, 0.5, 0.0]]

Rik
  • 467
  • 4
  • 14
  • How about print some variable at each iteration like `cross_entropy` or `ytf` to see how is the network converging? (A random thought: if the dataset is all 0's at the beginning and then all 1's, it might 'forget' about 0) – Blownhither Ma Jan 19 '18 at 16:31
  • Possible duplicate of [Tensorflow sigmoid and cross entropy vs sigmoid\_cross\_entropy\_with\_logits](https://stackoverflow.com/questions/46291253/tensorflow-sigmoid-and-cross-entropy-vs-sigmoid-cross-entropy-with-logits) – Maxim Jan 19 '18 at 18:52

1 Answers1

2

You are applying a categorical cross entropy function to binary classification

y_ * tf.log(ytf)

Here you have tf.log(ytf), so if ytf is 1, then tf.log(ytf) always evaluates to 0. So your function will never learn.

You need to use binary cross entropy function

y_ * tf.log(ytf) + (1. - y_) * tf.log(1. - ytf)

While training, always get the loss as output and keep printing once a while, it will help you debug your model much faster.

_, training_loss = sess.run([train_step, cross_entropy], feed_dict={xtf: x_batch, y_: y_batch})
layog
  • 4,661
  • 1
  • 28
  • 30
  • Now it is giving all values as nan. – Rik Jan 19 '18 at 16:49
  • Clip your `ytf` to some values near `0` and `1` as `tf.clip_by_value(ytf, 0.02, 0.98)` – layog Jan 19 '18 at 16:51
  • Its still not working. I have included my training data if you'd like to check. – Rik Jan 19 '18 at 17:54
  • print your `loss` with training and report the loss. Is it `0` or `nan`? – layog Jan 19 '18 at 18:10
  • The loss is nan. – Rik Jan 19 '18 at 18:19
  • Remove `tf.reduce_sum` and add `axis=0` to `tf.reduce_mean` in your loss function. It should not change your loss, it's just removing redundant operations from your loss, but they might be causing some problem. I'll run it myself and will revert – layog Jan 19 '18 at 18:35
  • Let us [continue this discussion in chat](http://chat.stackoverflow.com/rooms/163526/discussion-between-rik-and-layog). – Rik Jan 20 '18 at 06:10