Find Jobs
Hire Freelancers

Custom neural network training software

$1500-3000 USD

Fechado
Publicado há mais de 7 anos

$1500-3000 USD

Pago na entrega
I want a custom neural network training software which will need to be in GUI. I will use it to train thousands of models from hundreds of csv files, each containing a couple thousand inputs (arranged in time series). I will use this program to monitor and select the best model(s) for each corresponding csv file. Once the training is done, I also need the program to show the results using graphs/statistics (r^2,MSE,mean,SD,variance,histogram,etc). The training algorithm should use the Levenberg–Marquardt algorithm (feed-forward backprop) and the adaptive function should use the Gradient descent with momentum backpropagation. The ratio of train/validation/test data should also be customizable. I will need the program to generate a large amount of models for each csv file based on the following commands: 1. Input. This may span across several csv files. Number of input may fall within a range, and given a step size. (ie. 5 inputs per day, stretching 10-100 days backward, step size = 10 day.) 2. Training parameters/number of outputs. This may be a function of the inputs. (ie. IF input between tomorrow and [tomorrow+X] days increases more than Y times, THEN 1, ELSE 0 {where X and Y are functions of inputs} ) 3. Range of hidden layers (ie. 1-2) 4. Range of hidden nodes and step size Example of a command: Nodes 20-60 for each available layer, step size = 20. Example output: 1 Layer {(20),(40),(60)} 2 Layers {(20,0),(40,0),(60,0),(20,20),(40,20),(60,20),....(60,60)} 5. Transfer function (Hyperbolic tangent sigmoid, Log-sigmoid, or linear) Using the examples above, Command 1. will have 10 combinations (10, 20, 30, 40, 50, 60, 70, 80, 90, 100 days) Command. 2. will have 1 combination. Command 3. will have 2 combinations (1 hidden layer, 2 hidden layer) Command 4 will have 18 combinations {(20),(40),(60)}u{(20,0),(40,0),(60,0),(20,20),(40,20),(60,20),....(60,60)} Command no 5. Should have 3 combinations Therefore, for each csv file, there should be 10 x 1 x 2 x 18 x 3 = 1080 models. That’s about it. If possible, I would like the software to be able to run on multiple GPUs, if not that’s fine.
ID do Projeto: 12551109

Sobre o projeto

8 propostas
Projeto remoto
Ativo há 7 anos

Quer ganhar algum dinheiro?

Benefícios de ofertar no Freelancer

Defina seu orçamento e seu prazo
Seja pago pelo seu trabalho
Descreva sua proposta
É grátis para se inscrever e fazer ofertas em trabalhos
8 freelancers estão ofertando em média $2.915 USD for esse trabalho
Avatar do Usuário
Hi I am a very experienced statistician and programmer. I have completed several PhD level thesis projects involving advanced statistical analysis of data. I have worked with data from several companies and have done projects involving high level quantitative analysis and data interpretation skills to study the trends, time behaviour and compare the variables in the data. I can do advanced level analysis in SPSS, R, WEKA and excel tools like machine learning, hypothesis testing, forecasting, T-test, ANOVA etc. Looking forward to discussion, Best Regards, Suyash
$5.000 USD em 50 dias
5,0 (5 avaliações)
3,8
3,8
Avatar do Usuário
Dear, I have already read your job post in carefully and understood it well. I have many experiences on Machine Learning using neural network, SVM, Boosting, Adaboost, etc for 7+ years. And I have good and full experiences on mathematics and statistics, computation, numerical analysis, digital image processing using matlab for 10+ years and I had worked on many matlab project before. So, I can complete your project within your timeline, finely. Please let me know about your project detail and inform me. I am waiting for your good reply. Thanks and regards, Xie
$1.500 USD em 30 dias
5,0 (6 avaliações)
3,1
3,1
Avatar do Usuário
Hello, I have a strong inclination towards Programming and Machine learning and their applications in real life scenarios. A brief background of my skills and experience is outlined below. I have expertise in C-programming, Python, C++, Data Mining. Also I am interested in competitive programming to make the algorithms more efficient. Till now I have complete understanding of the topics covered in Data structures like linked lists, arrays, stacks , queues, dynamic programming, greedy method, trees etc and in machine learning I have the complete knowledge of Linear&Logistic Regression, Classification, Neural Networks, SVM, k-means algorithm, kernels, decision trees, genetic algorithms, rough sets, fuzzy logics etc. in machine learning. I have implemented these algorithms using R Programming Language as well. I have also done a project based on the above mentioned skill sets in which i had a data of a mobile company in that data i had to developed a prediction model which would be easily able to predict that whether a new customer that is taking the sim card of that company will switch to another network or not given a feature set of the customer after cleaning and pre-processing of the given data. I am looking for an opportunity to work with you just let me know if there is a project in which I could show-case my skill sets.
$2.500 USD em 30 dias
5,0 (4 avaliações)
2,5
2,5
Avatar do Usuário
Hello, I lead a US company specialized in machine learning and software development. We are a team of data scientists, AI researchers and software engineers. I am an expert in machine learning And predictive modeling. I have a PhD in physics and extensive experience in machine learning applications. We have experience in algorithmic trading, social networks analytics, recommender systems, NLP, image and pattern recognition, and predictive modeling in various fields. We are proficient in R, Python, Tensorflow, AML, C#, .NET, Angular, d3, Django, Quantopian, SQL, mongoDB. We are experts in neural nets, deep learning and machine learning applications.
$2.600 USD em 20 dias
0,0 (0 avaliações)
0,0
0,0
Avatar do Usuário
I have previously created a neural network for stock prices time series prediction. It took the past closing price as an input, used two hidden layers with 10 nodes in each layer, and would predict the tomorrow's price. I used gradient descent and backpropogation for determining the weights. (Not so related, but I have also used the combination of statistical ARIMA and neural networks for time series prediction as part of my phd thesis dissertation). I can do your project, please let me know if interested.
$3.000 USD em 30 dias
0,0 (0 avaliações)
0,0
0,0

Sobre o cliente

Bandeira do(a) THAILAND
Thailand
0,0
0
Membro desde dez. 19, 2016

Verificação do Cliente

Obrigado! Te enviamos um link por e-mail para que você possa reivindicar seu crédito gratuito.
Algo deu errado ao enviar seu e-mail. Por favor, tente novamente.
Usuários Registrados Total de Trabalhos Publicados
Freelancer ® is a registered Trademark of Freelancer Technology Pty Limited (ACN 142 189 759)
Copyright © 2024 Freelancer Technology Pty Limited (ACN 142 189 759)
Carregando pré-visualização
Permissão concedida para Geolocalização.
Sua sessão expirou e você foi desconectado. Por favor, faça login novamente.