pairwise and triplet data preparation

[yosajka]

Hi, thank you for your awesome work. I'm learning pytorch so it's little hard for me to understand your code. How do you prepare pairwise or triplet data and feed them into the model in the training phase?

[swuxyj]

torch.unsqueeze:

• Returns a new tensor with a dimension of size one inserted at the specified position.
• The returned tensor shares the same underlying data with this tensor.

Example:

>>> x = torch.tensor([1, 2, 3, 4])
>>> torch.unsqueeze(x, 0)
tensor([[ 1,  2,  3,  4]])
>>> torch.unsqueeze(x, 1)
tensor([[ 1],
        [ 2],
        [ 3],
        [ 4]])

pairwise

dist = (u.unsqueeze(1) - self.U.unsqueeze(0)).pow(2).sum(dim=2)

triplet

loss1 = 0
for row in range(s.shape[0]):
    # if has positive pairs and negative pairs
    if s[row].sum() != 0 and (~s[row]).sum() != 0:
        count += 1
        theta_positive = inner_product[row][s[row] == 1]
        theta_negative = inner_product[row][s[row] == 0]
        triple = (theta_positive.unsqueeze(1) - theta_negative.unsqueeze(0) - config["alpha"]).clamp(min=-100,
                                                                                                     max=50)
        loss1 += -(triple - torch.log(1 + torch.exp(triple))).mean()

[yosajka]

Thanks for your quick reply. Helped me a lot. Cheer!