<03> GRADIENT PASTLASH

GRADIENT PASTLASH (DESCENT)

Gradient Descent algoritmi weight "w" ning optimal qiymatini topishda qo'llaniladigan algoritmdir. Sodda qilib aytganda, bu algoritm bizlarga avtomatik tarzda optimal yechimni topib beradi. Ushbu algortim nafaqat w ning qiymatini topishda balki keyinchalik cost function(kelasi darslarimizda o'rganamiz) ni hisoblashda ham qo'llaniladi.

MATEMATIK FORMULASI (LOSS UCHUN)

AMALIYOT

# Training Data(O'rgatishdagi ma'lumotlar)
x_soat = [1.0, 2.0, 3.0]
y_baho = [2.0, 4.0, 6.0]

w = 1.0  #w uchun dastalbki taxminiy qiymat


# (Modelimiz)To'g'ri hisoblash uchun funksiya
def forward(x):
    return x * w


# Xatolik (Loss) ning funkisyasi
def loss(x, y):
    y_pred = forward(x)
    return (y_pred - y) * (y_pred - y)


# Gradient uchun funksiya
def gradient(x, y):  # d_loss/d_w
    return 2 * x * (x * w - y)


# Training dan avval
print("Bashorat (training dan avval)",  "4 soat o'qilganda:", forward(4))

# Training zanjiri (loop)
learning_rate =0.01
for epoch in range(10):
    for x_hb_qiym, y_hb_qiym in zip(x_soat, y_baho):
        # Hosilani hisoblash
        # w ning qiymatini yangilash
        # xatolikni hisoblab progressni chop qilish
        grad = gradient(x_hb_qiym, y_hb_qiym)
        w = w - learning_rate * grad
        print("\tgrad: ", x_hb_qiym, y_hb_qiym, round(grad, 2))
        l = loss(x_hb_qiym, y_hb_qiym)
    print("progress:", epoch, "w=", round(w, 2), "loss=", round(l, 2))

# Traningdan so'ng
print("Bashorat (training dan keyin)",  "4 saot o'qilganda: ", forward(4))

MATERIALLAR

604KB

3-lecture(Gradient Descent).pdf

pdf