Giter VIP home page Giter VIP logo

Comments (4)

magehrig avatar magehrig commented on June 16, 2024

Hi @Jee-King

Yes, this is correct. I think it would be better to instead sum up the number of events per time index instead. Of course, then the SNN pipeline must support this and I don't remember if Slayer does right now.

from snn_angular_velocity.

Jee-King avatar Jee-King commented on June 16, 2024

Hi, thanks for your reply!

I am reproducing the training process according to the paper. I use the MSE loss function instead of the function described in the paper. But I got a bad performance. Does the MSE loss function have a decisive effect on the results? By the way, could you provide the source code of the loss function in the paper?

from snn_angular_velocity.

magehrig avatar magehrig commented on June 16, 2024

I don't think that you will see a big difference if you use a slightly different loss. The challenge of training this SNN lies mostly in making sure that you have gradients for all layers. I used the following loss:

class RMSELoss:
    def __init__(self):
        self.str_to_idx = {
            'x': 0,
            'y': 1,
            'z': 2,
        }

    def compute(self, input_: torch.Tensor, target: torch.Tensor, time_start_idx: int=None, axis: str=None):
        assert len(input_.shape) == 3
        assert len(target.shape) == 3
        assert input_.shape == target.shape
        assert input_.shape[1] == 3
        """
        :param input: tensor of shape (batch, 3, time)
        :param target: tensor of shape (batch, 3, time)
        :param time_start_idx: Time-index from which to start computing the loss
        :param axis: {'x', 'y', 'z'}
        :return: Root mean squared error
        """
        if axis:
            axis_idx = self.str_to_idx[axis]
            input_ = input_[:, axis_idx, :]
            target = target[:, axis_idx, :]

        if time_start_idx:
            input_ = input_[..., time_start_idx:]
            target = target[..., time_start_idx:]

        return torch.mean(torch.sqrt(torch.mean((input_ - target) ** 2, dim=1)))

from snn_angular_velocity.

Jee-King avatar Jee-King commented on June 16, 2024

Thank you very much for your help!

from snn_angular_velocity.

Related Issues (8)

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.