### optical chemical structure recognition from images

I want to recognize name of the chemical structure from the image of chemical structure, like in the given image it is benzene structure I want to recognize that it is benzene from the image(I should be able to recognize all these structures as benzene) What way I could follow to achieve this? (like this…

### Why do current models use multiple normalization layers?

In most current models, the normalization layer is applied after each convolution layer. Many models use block: Conv->BatchNorm->ReLU repeatedly. But why do we need multiple BatchNorm layers? If we have a Conv that receives a normalized input, shouldn’t it spit out a normalized output? Isn’t it enough to place normalization layers only at the beginning…

### Choosing a framework / starting point

I’m looking for guidance in choosing a framework / starting point. I have a quite specific project in mind, but very little hands-on experience with ML / AI yet. I want to train a network with video data and have it transform pixel values over time on an input video. This is for an art…

### How does transformer leverage GPU which trains faster than RNN?

How does transformer leverage GPU which trains faster than RNN? I understand the parameter space of the transformer might be significantly larger than that of the RNN. But why does the transformer structure can leverage multi-gpu and accelerates its training?

### Should I consider mean or sampled value for action selection in ppo algorithm?

When considering the policy network in PPO algorithm, we need to fit a gaussian distribution to the neural network output(for a continuous action space problem). When I use this network to obtain action, should I sample from the fitted distribution or directly consider the mean output from the policy nn? Also, what should be the…

### What are some common heuristics that might be innate?

Here’s a question I might ask an AI to solve: “Colour the states of the USA using just 4 colours”. Now, a common heuristic a human might use is to start at one state and “work their way out”. Or start at an edge state. Now this seems to work best rather than just colouring…

### Is there an algorithm for “contextual recognition” with probabilities?

Example 1 An object is composed of 3 sub-objects. Object 1: 90% looks like an eye 10% looks like a wheel Object 2: 50% looks like an eye 50% looks like a wheel Object 3: 90% looks like a mouth 10% looks like a roof OK. So now we want to determine what the whole-object…

### How can an AI freely make decisions on a network?

Say a Deep Neural Net is created using Keras or Tensorflow. Usually when you want to make a prediction the user would invoke model.predict…. However, how would the actual AI system proactively invoke their own actions? Happy to clarify the question.

### What would be the implications of mistakenly adding bias after the activation function?

I was looking at the source code for a personal project neural network implementation, and the bias for each node was mistakenly applied after the activation function. The output of each node was therefore $\sigma\big(\sum_{i=1}^n w_i x_i\big)+b$ instead of $\sigma\big(\sum_{i=1}^n w_i x_i + b\big)$. Assuming standard back-propagation algorithms are used for training, what (if any)…

### Tabular Datasets where Deep neural networks outperforms XGBoost

Are there (complex) tabular datasets where deep neural networks (e.g. more than 3 layers) outperform traditional methods such as XGBoost by a large margin? I’d prefer tabular datasets rather than image datasets, since most image dataset are either too simple that even XGBoost can perform well (e.g. MNIST), or too difficult for XGBoost that its…