Stack Exchange Network

Stack Exchange network consists of 174 Q&A communities including Stack Overflow, the largest, most trusted online community for developers to learn, share their knowledge, and build their careers.

Visit Stack Exchange

  7. current community

    your communities

    more stack exchange communities

    company blog
Cross Validated

Questions tagged [transfer-learning]

A setting in machine learning when a model trained in one context/domain should then be applied to a different (but related) context/domain.

60 questions
info Newest Frequent Votes Active Unanswered
0
votes
0answers
16 views

segmentation and recognize multicolor background letters [closed]

I working on a car license plate detector. so I detect and crop the plate from images. my images don't have labels so I prefer using a pre-trained model that can recognize the plate numbers and ...
2
votes
0answers
22 views

How to Fine Tune a pre-trained network

I'm looking into using Transfer Learning to take the ResNet50 model trained on ImageNet and fine tune it to my own dataset using Keras. However, I feel I have some misconception about what exactly ...
1
vote
0answers
13 views

compute the KL divergence between two datasets

I have two datasets $D1$ and $D2$ in two different feature spaces $\mathcal{X}_{1} \in \Re^{m}$ and $\mathcal{X}_{2} \in \Re^{n}$. Further assume that the datasets have different number of data points....
0
votes
0answers
5 views

Will merging class labels in pre-trained model hurt transfer learning?

I'm using a large image dataset labeled with 15 classes to train a ConvNet model. The resulting model will then be used to enable transfer learning in a tiny dataset labeled with only 3 classes. The 3 ...
0
votes
0answers
8 views

Would training 1024x1024 images with pretrained resnet (224x224) be appropriate?

I want to use Resnet50 (or 101, or 152..) backbone for a segmentation task. My problem requires a lot of context, hence tiling the large image into 224x224 defeats the purpose. I was wondering if I ...
1
vote
0answers
16 views

Using PCA in domain adaptation

In literature, I see people using (Kernelized) Principle Component Analysis, not for feature extraction, but for domain adaptation. In other words, I have data from a source domain and I would like to ...
0
votes
1answer
36 views

What does a “similar” dataset mean in the context of fine tuning a CNN?

In https://arxiv.org/pdf/1809.09529.pdf it is said If the new dataset is similar to original dataset, we expect higher-level features in the CNN to be relevant to this dataset. Thus, it is ...
3
votes
1answer
76 views

How should I standardize input when fine-tuning a CNN?

I am working on a model for binary classification of skin samples from https://www.isic-archive.com as either benign or malignant. I want to use the VGG16 model pre-trained on ImageNet and fine-tune ...
0
votes
1answer
16 views

Should I be stepping down high a dimensional embedding when predicting low dimensional output

I'm using a ResNet-50 pretrained on ImageNet as a starting point for various image classifiers. Because the pretrained model has 1001 outputs, I have added a single dense layer with output size 500 ...
0
votes
1answer
44 views

using (deep) neural networks for a severely imbalanced image dataset when some classes have <10 images

Taking a long shot here. So I have a a small dataset of ~500 images with discrete labels from 1 to 9. My task is to detect the per-class and overall accuracy of this classification method using a (...
0
votes
1answer
133 views

Is Gradient Boosting Regression Tree able to learn linear models

Assume $Y$ is a linear function of a vector of variables $X$ (plus a noise term). The train data consists of ($X,Y$) such that $X \in ［0,1］$. Assume one use gbdt to learn this linear model. And if ...
0
votes
0answers
22 views

FineTuned VGG16 achieving great results on epoch 1, is this normal?

I'm training a model to classify whether a person is smiling (showing teeth) or not. I'm using Keras and I trained a VGG16 model loaded with the ImageNet weights, froze the first 4 layers and added a ...
0
votes
1answer
20 views

How do we call when one model is trained by another?

Suppose I have one model (say, for image classification) as black box. I don't posess it and don't know it's parameters. Suppose it is web API. Then I take bunch of images, classify them by this ...
0
votes
0answers
5 views

doubt regarding structural correspondence learning

I was going through the paper titled "Domain Adaptation with Structural Correspondence Learning" (http://john.blitzer.com/papers/emnlp06.pdf). I am writing this to resolve one doubt that I ran into. ...
-1
votes
2answers
39 views

Transfer learning scenario [closed]

I have an interesting question and I need some help. Consider the following problem. Let's suppose that I am doing regression with neural networks. As input I have a set of measurements, arrays of $N$ ...
1
vote
0answers
48 views

Transfer Learning for Multivariate Regression

As I understand it, transfer learning is termed as using the parameters of a pre-trained model, which was initially trained on a particular 'source' task, and have it train on another related 'target' ...
0
votes
1answer
194 views

Transfer learning for audio

I know that when working with images, what people normally do is download a big model trained with huge data and freeze most of the layers except the lasts ones to train them with their own data. I'm ...
4
votes
1answer
422 views

Transfer learning: How and why retrain only final layers of a network?

In this video, Prof. Andrew Ng says regarding transfer learning: Depending on how much data you have, you might just retrain the new layers of the network, or maybe you could retrain even more ...
1
vote
0answers
277 views

Keras K.learning_phase() [closed]

I've tried to make transfer learning via keras.applications like here (https://keras.io/applications/) for binary classification of images (crocodiles and clocks). load model without top layers add ...
0
votes
0answers
374 views

Transfer learning and Generative adversarial networks

I need to do image classification. I have small data, so I need to use Transfer learning. And also, I have some negative samples on my data, so I will use Gans. How can I combine GANs and transfer ...
1
vote
1answer
554 views

Fine Tuning vs. Transferlearning vs. Learning from scratch

In my master thesis, I am researching on transfer learning on a specific use Case, a traffic sign detector implemented as a Single Shot Detector with a VGG16 base network for classification. The ...
1
vote
0answers
24 views

Transfer learning for image classification

When working with transfer learning for image classification, I would like to freeze only a part of the convolutional base of a pretrained model while adding a classifier (some shallow network) on top ...
0
votes
0answers
10 views

How to distinguish whether a predicted result belongs to a new class or existing class in zero-shot learning?

In zero-shot learning, a system can predict a class even the system uses unseen data in training data. In general classification tasks, the number of labels is fixed. So I just need to pick a class ...
0
votes
1answer
53 views

Adapt speech recognition for Shakespeare english

We need to be able to search the works of Shakespeare by voice. The way I see it, the goal is if I quote into the microphone: "Yet but three come one more. Two of both kinds make up four. Ere ...
5
votes
1answer
2k views

Transfer Learning on generative adversarial networks (GANs)

Is it possible to apply transfer learning to GANs and if so, what are some examples of someone having tried this?
0
votes
0answers
20 views

Intuition behind attribute learning in machine learning

I am working on zero shot learning, a machine paradigm that allows the prediction of classes even when no samples of the class were applicable during training. In this paradigm, there is a lot of hype ...
0
votes
0answers
1k views

How to Train LSTM across Multiple Time Series datasets

I believe there are generalizable mapping rules that can be extracted from multiple time series data-sets I have. Each data-set represents a specific company, and while I am able to train on each ...
1
vote
1answer
99 views

Comparing distribution of training and test dataset

When we fit a model to a training dataset, our basic assumption is that the test dataset will also come from the same (or similar) distribution. How to check if the distribution of the training and ...
9
votes
2answers
3k views

Where to find pre-trained models for transfer learning [closed]

I am new to the machine learning field, but I wanted to try and implement a simple classification algorithm with Keras. Unfortunately, I have a very small set of data, so I thought to try to apply ...
0
votes
1answer
74 views

Can object detection be done by passing parts of images to an image classification network? [closed]

I am using a pre-trained image classification network. How is the object detection performace by passing the parts of the image individually to detect various objects. is there any other way to ...
26
votes
6answers
6k views

Why study convex optimization for theoretical machine learning?

I am working on theoretical machine learning — on transfer learning, to be specific — for my Ph.D. Out of curiosity, why should I take a course on convex optimization? What take-aways from convex ...
4
votes
1answer
498 views

how is covariate shift associated with domain adaptation?

This question is in context to transfer learning. During my presentation on transfer learning, I was asked the difference between covariate shift and domain adaptation. All I know is that covariate ...
0
votes
0answers
358 views

Keyword Spotting: How to train a model with general speech corpus?

I am trying to find a correct way to train a DNN based keyword spotting (Deep KWS) with general speech corpus (VS data) described in this paper (Chen, Guoguo, Carolina Parada, and Georg Heigold. "...
1
vote
0answers
11 views

physical significance of attribute learning

Just reading an article on attribute learning and its usage in transfer learning and novelty detection. I have one doubt pertaining to this phenomenon. What is the outcome of attribute learning ? Are ...
3
votes
2answers
558 views

Generate images from CNN model built with transfer learning

I have trained an image classifier (for my own face) via the transfer learning method, in two ways (in order to compare efficiencies later on): 1- I created a ...
1
vote
1answer
30 views

Robust machine learning for slightly different class proportions in multiple data sets

Say we have n similar data sets, with the same variables, and outcome labels x and y. In these data sets, domains slightly differ as suggested by the proportion of the minority class x (ranging from 1%...
1
vote
0answers
39 views

marginal distribution in context to tranfer learning

In transfer learning, we often come across marginal distributions in the target and source domain as being different. In simple terms, what does it essentially mean? I have read about marginal ...
1
vote
0answers
48 views

Forecasting monthly demand given yearly shocks

I have a dataset of monthly demand in dollars. We'd like to forecast the next year by month using an ARIMA technique. We have a pool of 500 clients who submit invoices on behalf of hundreds of ...
5
votes
0answers
4k views

Change image input size of a pre-trained convnet

maybe this question will sound a bit as a newbie one but I'd like to have some clarification. I'm using a VGG16-like convnet, pre-trained with VGG16 weights and edited top layers to work with my ...
1
vote
0answers
196 views

How much data is needed for transfer learning?

I have a pre-trained sentiment classifier on Amazon reviews and I want to use it for transfer learning to classify sentiments on a specific aspect of some products, for example how a person feels ...
3
votes
0answers
2k views

Transfer learning on faster rcnn and tensorflow

I am trying to do transfer learning to reuse a pretrained neural net. I got the tensorflow faster rcnn official example to work, and now i would like to reuse it to detect my own classes. This is the ...
1
vote
0answers
23 views

difference between re-training initial layers and final layers of NN

I was going through a brief tutorial on "transfer learning" available on https://www.analyticsvidhya.com/blog/2017/06/transfer-learning-the-art-of-fine-tuning-a-pre-trained-model/ In this blog, a ...
1
vote
0answers
68 views

Training a model good enough for transfer learning

I am working on transfer learning for a robotic application. Right now, we have want to train a deep learning model (convolution network) using our current dataset (source task), and then use the pre-...
1
vote
1answer
241 views

Can I use transfer learning on a 3 channel conv net (ie VVG16) for a 1 channel classification problem?

I am trying to create a convolutional neural network for multilabel classification of greyscale images showing the outline of people. I do not have much available data and overfitting is a problem. ...
2
votes
1answer
427 views

Fine-tuning with a subset of the same data

My understanding of fine-tuning is to take a pre-trained model trained on a similar but separate dataset and update the weights of a portion of the model on your dataset. I'm not sure if this is ...
4
votes
1answer
228 views

Modifying ordinary least squares (OLS) in ridge regression to perform transfer learning

I have a question on using ridge regression for transfer learning. Transfer learning is a type of Machine Learning where knowledge from the source domain when performing a task is transfered to the ...
0
votes
0answers
489 views

Tensorflow object detection : Using transfer learning on local running

From Tensorflow docs, we can use transfer learning for object detection when you run from cloud. Also, can we using transfer-learning for running locally ? i see that we have a doc about running on ...
3
votes
0answers
1k views

Is there a way to merge two trained neural networks?

Lets say I pick some network layout (recurrent and/or deep is fine if it matters I'm interested to know why), then make two neural networks A and B using that layout that are initially identical. Now ...
2
votes
0answers
266 views

Different sized inputs for batch training in fully CNNs

The idea of transfer learning is to use already trained networks for another purposes to the one it was initially trained for. Using fully convolutional networks, the activation maps can be used to ...
5
votes
2answers
381 views

Python package that allows to train a CRF on two datasets

I am looking for a Python package that allows to train a conditional random field (CRF) on two datasets. For example: I have two datasets, dataset A and dataset B. I want to train a conditional ...

1 2