Closed. This question does not meet Stack Overflow guidelines. It is not currently accepting answers.
This question does not appear to be about programming within the scope defined in the help center.
Closed 3 years ago.
Improve this question
Prior to 2017, it was relatively simple to understand which CNN was the best to classify images with the imagnet yearly competition.
In 2017 the imagenet competition was divided into different tasks with winners such as this. In 2018, the competition moved to kaggle and became about 3D detection.
I am interested in image classification only and there no longer seems to be a competition for this.
Does anyone know what neural network was recognised as the best for image classification in 2018?
If i recall correct I think it is Googles NasNet. It's a very cool (and computer intensive) method used to design the model architecture, but good for transfer learning and prediction. I would recommend taking a look at the NasNet-paper
It should also be available to use through keras.application
This is a really good question. I was wondering about the same and played around with some of the models that are on TensorFlow Hub. So, here are my two cents.
The current best models in terms of performance on ImageNet are the ones which are obtained with Progressive Neural Architecture Search. On the other hand, these models are incredibly slow to train because they are huge. When it comes to the models such as InceptionNet, ResNet, and VGG, this is a good link to check out the performance compared to the training/inference speed.
My personal experience is that if you want to maximize performance, use ResNet152. If you want a relatively fast CNN, while achieving good performance, go with ResNet50. When it comes to the VGG nets, I played around with the TF-Slim implementation but it was slower than ResNet50, with performance around the same. Finally, I can't say much about Inception because I didn't use it. In the end, I went with ResNet152, because it yield the best performance for me (Please note that I was using a pre-trained version and I was fine-tuning it to my task).
To summarize, I think that there is no general best CNN. I would avoid using VGG16/19, because it yields worse performance than ResNet50, while being slower. If you have access to a lot of computational power, go with Resnet152 or PNASNet. Again, this my opinion based on my personal experience by playing around with the pre-trained models on TF-Hub.
Related
Closed. This question does not meet Stack Overflow guidelines. It is not currently accepting answers.
This question does not appear to be about programming within the scope defined in the help center.
Closed 1 year ago.
Improve this question
I have using Neural Network for a classification problem and I am now at the point to tune all the hyperparameters.
For now, I saw many different hyperparameters that I have to tune :
Learning rate
batch-size
number of iterations (epoch)
For now, my tuning is quite "manual" and I am not sure I am not doing everything in a proper way. Is there a special order to tune the parameters? E.g learning rate first, then batch size, then ... I am not sure that all these parameters are independent. Which ones are clearly independent and which ones are clearly not independent? Should we then tune them together? Is there any paper or article which talks about properly tuning all the parameters in a special order?
There is even more than that! E.g. the number of layers, the number of neurons per layer, which optimizer to chose, etc...
So the real work in training a neural network is actually finding the best-suited parameters.
I would say there is no clear guideline because training a machine learning algorithm, in general, is always task-specific.
You see, there are many hyperparameters to tune, and you won't have time to try out every combination of each. For many hyperparameters, you will build somewhat of intuition on what a good choice would be, but for now, a great starting point is always using what has been proven by others to work. So if you find a paper on the same or similar task you could try to use the same or similar parameters as them too.
Just to share with you some small experiences I've made:
I rarely vary the learning rate. I mostly choose the Adam optimizer and stick with it.
The batch size I try to choose as big as possible without running out of memory
number of iterations you could just set to e.g. 1000. You can always look at the current loss and decide for yourself if you can stop when the net e.g. isn't learning anymore.
Keep in mind these are in no way rules or strict guidelines. Just some ideas until you've got a better intuition yourself. The more papers you've read and more nets you've trained you will understand what to chose when better.
Hope this serves a good starting point at least.
Closed. This question needs to be more focused. It is not currently accepting answers.
Want to improve this question? Update the question so it focuses on one problem only by editing this post.
Closed 5 years ago.
Improve this question
I'm trying to understand if the project I'm thinking about is feasible or not using Neural Networks. I'm aware of apps like MakeApp and FakeApp which use neural networks to manipulate human faces.
My question is - Can modern (2018) neural networks be trained to identify aspects of human facial attractiveness and give a percentile score?
For example, given an image, I want to know if the neural network thinks this image is in the top 20% facial attractiveness. If possible, how big of a dataset I need to be able to train such network? Is it tens of thousands of human-scored images?
Certainly. There is already research being done on developing deep learning / convolutional neural networks to do exactly this. Four recent references as of January 2018 are given below.
The main challenges with doing it are:
Acquiring a large enough dataset (human face images and their respective attractiveness scores) with proper subject approval.
The fact that attractiveness is subjective and varies with ethnic group and culture. Therefore such training data will have a broader range of labels than in more classical recognition tasks such as object detection (for which the label is binary), leading to more uncertainty in the network's predictions. For this reason most research focuses on training networks for a specific group.
This research area isn't being developed hugely (at least in academia) at the moment most likely because of ethical considerations with acquiring such sensitive data and dubious uses. I suspect that now companies like OKCupid and Match.com are or will be developing this research privately for the purposes of automatic match making.
Xu et al., A new humanlike facial attractiveness predictor with cascaded fine-tuning deep learning model, arXiv 2015,
paper
Gan et al., Deep self-taught learning for facial beauty prediction, Neurocomputing 2014
paper
Wang et al., Attractive or Not?: Beauty Prediction with Attractiveness-Aware Encoders and Robust Late Fusion, ACM international conference on Multimedia 2014
paper
Shen et al., Fooling Neural Networks in Face Attractiveness Evaluation: Adversarial Examples with High Attractiveness Score But Low Subjective Score
Multimedia Big Data (BigMM), 2017 IEEE Third International Conference on
paper
Well I think this can be done. So first of all you need to specify the parameters for attractiveness. On what I have researched, I know 2 paarmeters that directly contribute to attractiveness are prominent jawline and cheekbones. I am sure that there are many more features that could be considered.But for the sake of examples lets take these two.
But you have to use a deep neural network. Since the different layers will contribute to simpler functions like getting the edges of face.
So the initial layers will get the edges, and after a few layers you will get the jawline and cheekbones and you can test them against your training set for attractiveness.
I am not sure how to get the training set. But you can use tinder to get images but scoring them would be an issue.
Nice idea and I hope that you could implement it for learning purpose.
Cheers.!!!
Closed. This question does not meet Stack Overflow guidelines. It is not currently accepting answers.
This question does not appear to be about programming within the scope defined in the help center.
Closed 2 years ago.
Improve this question
Can I use reinforcement learning on classification? Such as human activity recognition? And how?
There are two types of feedback. One is evaluative that is used in reinforcement learning method and second is instructive that is used in supervised learning mostly used for classification problems.
When supervised learning is used, the weights of the neural network are adjusted based on the information of the correct labels provided in the training dataset. So, on selecting a wrong class, the loss increases and weights are adjusted, so that for the input of that kind, this wrong class is not chosen again.
However, in reinforcement learning, the system explores all the possible actions, class labels for various inputs in this case and by evaluating the reward it decides what is right and what is wrong. It may be the case too that until it gets the correct class label it may be giving wrong class name as it is the best possible output it has found till now. So, it doesn't make use of the specific knowledge we have about the class labels, hence slows the convergence rate significantly as compared to supervised learning.
You can use reinforcement learning for classification problems but it won't be giving you any added benefit and instead slow down your convergence rate.
Short answer: Yes.
Detailed answer: yes but it's an overkill. Reinforcement learning is useful when you don't have labeled dataset to learn the correct policy, so you need to develop correct strategy based on the rewards. This also allows to backpropagate through non-differentiable blocks (which I suppose is not your case). The biggest drawback of reinforcement learning methods is that thay are typically took a VERY large amount of time to converge. So, if you possess labels, it would be a LOT more faster and easier to use regular supervised learning.
You may be able to develop an RL model that chooses which classifier to use. The gt labels being used to train the classifiers and the change in performance of those classifiers being the reward for the RL model. As others have said, it would probably take a very long time to converge, if it ever does. This idea may also require many tricks and tweaks to make it work. I would recommend searching for research papers on this topic.
Closed. This question is opinion-based. It is not currently accepting answers.
Want to improve this question? Update the question so it can be answered with facts and citations by editing this post.
Closed 4 years ago.
Improve this question
Let me just start by saying I only took the undergrad AI class at school so I know just enough to be dangerous.
Here's the problem I'm looking to solve...accurate credit scoring is a key part to the success of my business. Currently we rely on a team of actuaries and statistical analysis to suss out patterns in the few dozen variables we track about each individual that indicate that they may be a low or high credit risk. As I understand it this is exactly the type of job that neural nets are great at solving, that is, finding high order relationships across many inputs that a human would likely never spot and then rendering a decision or output that is on average more accurate than what a trained human could do. In short, I want to be able to input your name, address, marital status, what car you drive, where you work, hair color, favorite food, etc in and get a credit score back.
My question is what type or architecture for a neural network would be best for this particular problem. I've done a bit of research and it seems I'm generating questions faster than I'm finding answers at this point. The best I've been able to come up with is some kind of generative deep neural network with multiple hidden layers where each layer is able to abstract one level beyond the previous one. Im assuming it's going to be feed-forward just because it seems to be the default. We have historical data on all previous customers including the information we used to make the initial score as well as data on what type of credit risk they actually turned out to be. This would seem to lend itself to unsupervised learning. Where I'm lost is in number of layers, how the layers are different from each other, size of each layer, connectedness of each of the perceptrons and so on. The more I dig the more I'm getting into research papers that are over my head so I just need some smart person to point me in the right direction
Does anyone have any ideas? Again, I don't need a thorough explanation just a general area I should focus on.
This is supervised learning since you have actual data that can be labelled. It's also feedforward since you're not predicting time series but assigning scores. Further, you should probably just prepare your data (assigning credit scores manually or with some rough heuristic) and start experimenting with some tools before you invest time into implementing state-of-the-art architectures. A multi-layer-perceptron (MLP) with 1 hidden layer is a sufficient starting point for such a problem. From there on, you can train the network to generalize your credit assignment heuristic you began with.
You should know that most "new" architectures you probably read about while researching are dealing with much more difficult problems than credit scoring (speech/image/character recognition/detection). There is a collection of papers on the scenario of credit scoring / risk classification, so I'd recommend reshifting your focus from architectures to actual case studies (see e.g. this paper). Just pick a recent paper with MLPs and apply their parameters. Start simple and improve the system incrementally (as #roganjosh stated).
Closed. This question is off-topic. It is not currently accepting answers.
Want to improve this question? Update the question so it's on-topic for Stack Overflow.
Closed 11 years ago.
Improve this question
I am a computer science student and i have to choose the theme of my future research work. I really want to solve some scientific problems in chemistry(or maybe biology) using computers. Also I have huge interest in machine learning sphere.
I have been surfing over internet for a while, and have found some particular references on that kind of problems. But, unfortunately, that stuff is not enough for me.
So, I am interested in the Community's recommendation of particular resources that present the application of an ML technique to solve a problem in chemistry--e.g., a journal article or a good book describing typical (or the new ones) problems in chemistry being solved "in silico".
i should think that chemistry, as much as any domain, would have the richest supply of problems particularly suited for ML. The rubric of problems i have in mind are QSAR (quantitative structure-activity relationships) both for naturally occurring compounds and prospectively, e.g., drug design.
Perhaps have a look at AZOrange--an entire ML library built for the sole purpose of solving chemistry problems using ML techniques. In particular, AZOrange is a re-implementation of the highly-regarded GUI-driven ML Library, Orange, specifically for the solution of QSAR problems.
In addition, here are two particularly good ones--both published within the last year and in both, ML is at the heart (the link is to the article's page on the Journal of Chemoinformatics Site and includes the full text of each article):
AZOrange-High performance open source machine learning for QSAR modeling in a graphical programming environment.
2D-Qsar for 450 types of amino acid induction peptides with a novel substructure pair descriptor having wider scope
It seems to me that the general natural of QSAR problems are ideal for study by ML:
a highly non-linear relationship between the expectation variables
(e.g, "features") and the response variable (e.g., "class labels" or
"regression estimates")
at least for the larger molecules, the structure-activity
relationships is sufficiently complex that they are at least several
generations from solution by analytical means, so any hope of
accurate prediction of these relationships can only be reliably
performed by empirical techniques
oceans of training data pairing analysis of some form of
instrument-produced data (e.g., protein structure determined by x-ray
crystallography) with laboratory data recording the chemical behavior
behavior of that protein (e.g., reaction kinetics)
So here are a couple of suggestions for interesting and current areas of research at the ML-chemistry interface:
QSAR prediction applying current "best practices"; for instance, the technique that won the NetFlix Prize (awarded sept 2009) was not based on a state-of-the-art ML algorithm, instead it used kNN. The interesting aspects of the winning technique are:
the data imputation technique--the technique for re-generating the data rows having one or more feature missing; the particular
technique for solving this sparsity problem is usually referred to by
the term Positive Maximum Margin Matrix Factorization (or
Non-Negative Maximum Margin Matrix Factorization). Perhaps there are
a interesting QSAR problems which were deemed insoluble by ML
techniques because of poor data quality, in particular sparsity.
Armed with PMMMF, these might be good problems to revisit
algorithm combination--the rubric of post-processing techniques that involve combining the results of two or more
classifiers was generally known to ML practitioners prior to the
NetFlix Prize but in fact these techniques were rarely used. The most
widely used of these techniques are AdaBoost, Gradient Boosting, and
Bagging (bootstrap aggregation). I wonder if there are some QSAR
problems for which the state-of-the-art ML techniques have not quite
provided the resolution or prediction accuracy required by the
problem context; if so, it would certainly be interesting to know if
those results could be improved by combining classifiers. Aside from their often dramatic improvement on prediction accuracy, an additional advantage of these techniques is that many of them are very simple to implement. For instance, Bagging works like this: train your classifier for some number of epochs and look at the results; identify those data points in your training data that caused the poorest resolution by your classifier--i.e., the data points it consistently predicted incorrectly over many epochs; apply a higher weight to those training instances (i.e., penalize your classifier more heavily for an incorrect prediction) and re-train y our classifier with this "new" data set.