tra-analysis/data analysis/analysis/regression.py

218 lines
7.7 KiB
Python
Raw Normal View History

2019-09-26 18:35:37 +00:00
# Titan Robotics Team 2022: CUDA-based Regressions Module
# Written by Arthur Lu & Jacob Levine
# Notes:
2019-10-29 14:41:49 +00:00
# this should be imported as a python module using 'import regression'
# this should be included in the local directory or environment variable
# this module is cuda-optimized and vectorized (except for one small part)
2019-09-26 18:35:37 +00:00
# setup:
__version__ = "1.0.0.002"
2019-09-30 21:02:32 +00:00
# changelog should be viewed using print(regression.__changelog__)
2019-09-26 18:35:37 +00:00
__changelog__ = """
2019-11-05 19:45:35 +00:00
1.0.0.002:
-Added more parameters to log, exponential, polynomial
-Added SigmoidalRegKernelArthur, because Arthur apparently needs
to train the scaling and shifting of sigmoids
1.0.0.001:
-initial release, with linear, log, exponential, polynomial, and sigmoid kernels
-already vectorized (except for polynomial generation) and CUDA-optimized
2019-09-26 18:35:37 +00:00
"""
__author__ = (
"Jacob Levine <jlevine@imsa.edu>",
)
__all__ = [
'factorial',
'take_all_pwrs',
'num_poly_terms',
'set_device',
'LinearRegKernel',
'SigmoidalRegKernel',
'LogRegKernel',
'PolyRegKernel',
'ExpRegKernel',
'SigmoidalRegKernelArthur',
'SGDTrain',
'CustomTrain'
]
# imports (just one for now):
import torch
device = "cuda:0" if torch.torch.cuda.is_available() else "cpu"
#todo: document completely
def factorial(n):
if n==0:
return 1
else:
return n*factorial(n-1)
def num_poly_terms(num_vars, power):
if power == 0:
return 0
2019-09-27 14:48:05 +00:00
return int(factorial(num_vars+power-1) / factorial(power) / factorial(num_vars-1)) + num_poly_terms(num_vars, power-1)
2019-09-26 18:35:37 +00:00
def take_all_pwrs(vec,pwr):
#todo: vectorize (kinda)
combins=torch.combinations(vec, r=pwr, with_replacement=True)
out=torch.ones(combins.size()[0])
for i in torch.t(combins):
out *= i
return torch.cat(out,take_all_pwrs(vec, pwr-1))
def set_device(new_device):
global device
device=new_device
class LinearRegKernel():
parameters= []
weights=None
bias=None
def __init__(self, num_vars):
self.weights=torch.rand(num_vars, requires_grad=True, device=device)
self.bias=torch.rand(1, requires_grad=True, device=device)
self.parameters=[self.weights,self.bias]
def forward(self,mtx):
long_bias=self.bias.repeat([1,mtx.size()[1]])
return torch.matmul(self.weights,mtx)+long_bias
class SigmoidalRegKernel():
parameters= []
weights=None
bias=None
sigmoid=torch.nn.Sigmoid()
def __init__(self, num_vars):
self.weights=torch.rand(num_vars, requires_grad=True, device=device)
self.bias=torch.rand(1, requires_grad=True, device=device)
self.parameters=[self.weights,self.bias]
def forward(self,mtx):
long_bias=self.bias.repeat([1,mtx.size()[1]])
return self.sigmoid(torch.matmul(self.weights,mtx)+long_bias)
class SigmoidalRegKernelArthur():
parameters= []
weights=None
in_bias=None
scal_mult=None
out_bias=None
sigmoid=torch.nn.Sigmoid()
def __init__(self, num_vars):
self.weights=torch.rand(num_vars, requires_grad=True, device=device)
self.in_bias=torch.rand(1, requires_grad=True, device=device)
self.scal_mult=torch.rand(1, requires_grad=True, device=device)
2019-09-27 14:48:05 +00:00
self.out_bias=torch.rand(1, requires_grad=True, device=device)
2019-09-26 18:35:37 +00:00
self.parameters=[self.weights,self.in_bias, self.scal_mult, self.out_bias]
def forward(self,mtx):
long_in_bias=self.in_bias.repeat([1,mtx.size()[1]])
long_out_bias=self.out_bias.repeat([1,mtx.size()[1]])
2019-09-27 15:13:17 +00:00
return (self.scal_mult*self.sigmoid(torch.matmul(self.weights,mtx)+long_in_bias))+long_out_bias
2019-09-26 18:35:37 +00:00
class LogRegKernel():
parameters= []
weights=None
in_bias=None
scal_mult=None
out_bias=None
def __init__(self, num_vars):
self.weights=torch.rand(num_vars, requires_grad=True, device=device)
self.in_bias=torch.rand(1, requires_grad=True, device=device)
self.scal_mult=torch.rand(1, requires_grad=True, device=device)
2019-09-27 14:48:05 +00:00
self.out_bias=torch.rand(1, requires_grad=True, device=device)
2019-09-26 18:35:37 +00:00
self.parameters=[self.weights,self.in_bias, self.scal_mult, self.out_bias]
def forward(self,mtx):
long_in_bias=self.in_bias.repeat([1,mtx.size()[1]])
long_out_bias=self.out_bias.repeat([1,mtx.size()[1]])
2019-09-27 15:13:17 +00:00
return (self.scal_mult*torch.log(torch.matmul(self.weights,mtx)+long_in_bias))+long_out_bias
2019-09-26 18:35:37 +00:00
class ExpRegKernel():
parameters= []
weights=None
in_bias=None
scal_mult=None
out_bias=None
def __init__(self, num_vars):
self.weights=torch.rand(num_vars, requires_grad=True, device=device)
self.in_bias=torch.rand(1, requires_grad=True, device=device)
self.scal_mult=torch.rand(1, requires_grad=True, device=device)
2019-09-27 14:48:05 +00:00
self.out_bias=torch.rand(1, requires_grad=True, device=device)
2019-09-26 18:35:37 +00:00
self.parameters=[self.weights,self.in_bias, self.scal_mult, self.out_bias]
def forward(self,mtx):
long_in_bias=self.in_bias.repeat([1,mtx.size()[1]])
long_out_bias=self.out_bias.repeat([1,mtx.size()[1]])
2019-09-27 15:13:17 +00:00
return (self.scal_mult*torch.exp(torch.matmul(self.weights,mtx)+long_in_bias))+long_out_bias
2019-09-26 18:35:37 +00:00
class PolyRegKernel():
parameters= []
weights=None
bias=None
power=None
def __init__(self, num_vars, power):
self.power=power
num_terms=num_poly_terms(num_vars, power)
self.weights=torch.rand(num_terms, requires_grad=True, device=device)
self.bias=torch.rand(1, requires_grad=True, device=device)
self.parameters=[self.weights,self.bias]
def forward(self,mtx):
#TODO: Vectorize the last part
cols=[]
for i in torch.t(mtx):
cols.append(take_all_pwrs(i,self.power))
new_mtx=torch.t(torch.stack(cols))
long_bias=self.bias.repeat([1,mtx.size()[1]])
return torch.matmul(self.weights,new_mtx)+long_bias
def SGDTrain(kernel, data, ground, loss=torch.nn.MSELoss(), iterations=1000, learning_rate=.1, return_losses=False):
optim=torch.optim.SGD(kernel.parameters, lr=learning_rate)
data_cuda=data.to(device)
ground_cuda=ground.to(device)
if (return_losses):
losses=[]
for i in range(iterations):
with torch.set_grad_enabled(True):
optim.zero_grad()
pred=kernel.forward(data_cuda)
ls=loss(pred,ground_cuda)
losses.append(ls.item())
ls.backward()
optim.step()
return [kernel,losses]
else:
for i in range(iterations):
with torch.set_grad_enabled(True):
optim.zero_grad()
pred=kernel.forward(data_cuda)
ls=loss(pred,ground_cuda)
ls.backward()
optim.step()
return kernel
def CustomTrain(kernel, optim, data, ground, loss=torch.nn.MSELoss(), iterations=1000, return_losses=False):
data_cuda=data.to(device)
ground_cuda=ground.to(device)
if (return_losses):
losses=[]
for i in range(iterations):
with torch.set_grad_enabled(True):
optim.zero_grad()
pred=kernel.forward(data)
ls=loss(pred,ground)
losses.append(ls.item())
ls.backward()
optim.step()
return [kernel,losses]
else:
for i in range(iterations):
with torch.set_grad_enabled(True):
optim.zero_grad()
pred=kernel.forward(data_cuda)
ls=loss(pred,ground_cuda)
ls.backward()
optim.step()
return kernel