📖STUDY 웹 브라우저 속 머신러닝 Tensorflow.js | 로지스틱 회귀를 사용한 분류

import * as tf from "@tensorflow/tfjs";

const N = 100;
const c1 = tf.randomNormal([N, 2]).add([2.0, 1.0]);
const c2 = tf.randomNormal([N, 2]).add([-2.0, -1.0]);

const l1 = tf.ones([N, 1]);
const l2 = tf.ones([N, 1]);

const xs = c1.concat(c2);
const input = xs.concat([2 * N, 1], 1); // 1 as bias
const ys = l1.concat(l2);

const w = tf.randomNormal([3, 1]).sub(-0.5).variable();

const f_x = (x) => {
  return tf.sigmoid(x.matmul(w));
};

const loss = (pred, label) => {
  return tf.losses.sigmoidCrossEntropy(pred, label).asScalar();
};

const optimizer = tf.train.adam(0.07);
for (let i = 0; i < 100; i++) {
  const l = optimizer.minimize(() => loss(f_x(input), ys), true);
  losses.push(l.dataSync());
}

const model = tf.sequential();
model.add(tf.layers.dense(units: 1, batchInputShape: [null,2]));

const loss = (pred, label) => {
  return tf.losses.sigmoidCrossEntropy(pred, label).asScalar();
};

model.compile({
    loss: loss,
    optimizer: 'adam',
    metrics: ['accuracy']
});

async function training() {
    const history = await model.fit(xs, ys, {epochs: 100});
    ...
    model.predict(xs);
}

training();