As per your requirements, I would suggest that you start with any simple CNN network.
CNNs take advantage of the hierarchical pattern in data and assemble
more complex patterns using smaller and simpler patterns. Therefore,
on the scale of connectedness and complexity, CNNs are on the lower
Here is a Keras example:
model = models.Sequential()
However, if we can model the environment, why should we want to employ a model-free algorithm?
Depends what you mean by "model the environment". There are two kinds of model:
Distribution model, which provides full access to a function like $p(r,s'|s,a)$, the probability of observing reward $r$ and next state $s'$ given starting state $s$ and ...
You have not described exactly what the tasks will be, but there are some open source libraries for real time pose tracking. For example, OpenPose is one that can be configured to track the body, the hands and the face. However, this is only going to give you predicted pose information for each frame. If the subjects are meant to be doing specific tasks, e.g....
An ontology at its most abstract is a model of the world. It describes concepts that exist in the world and how those concepts are related.
Ontologies are similar to taxonomies. A taxonomy is a tree-like hierarchy that organizes concepts in increasing levels of specificity. What an ontology adds is a second type of link between those concepts that explains ...
4 years and 6 months have past since your unanswered question.
That is an eternity in terms of Machine Learning and things have evolved a lot. So I will answer about the present, not the past.
Today, there are some code generating models, like GitHub Copilot and OpenAi Codex which are based on NLG (Natural Language Generation). The principle is very simple:
You can find information similar to exposed by Neil, but with more theoretical detail, in the book Deep Learning (Goodfellow et al., 2016) in the chapter 10 (Recurrent networks), more specifically in 10.2.3 Recurrent Networks as Directed Graphical Models and other subchapters.
Additional, related with pointer networks there are people changing the LSTM with ...
I do not think there is one standard way to do this, it will depend too much on context. Ultimately you want the agent to output a stop action that is different from a continue action.
That stop/continue choice could either be part of the existing action encoding, additional data in parallel with the action sequence, or an entirely separate action choice on ...
What you are suggesting is similar to active learning and reward modelling.
To summarize both quickly, active learning is used when data are scarce or when the labeling process is too time consuming (almost always the case in NLP). To speed up the process, the idea is to train a model in performing not only a task, but also an estimation of its uncertainty ...
There is no right answer as it depends on many factors. But here are some keywords you can look into:
Keywords about your problem:
About the model, I'd recommend checking ARIMA. But before jumping into code.
A good problem solving with Data Science is a dynamic process for deeply understanding ...
Regarding your first point, it depends on what neural network you would like to use, the sensor temporal resolution, and the capabilities of the embedded system. You can figure out the number of operations required for a forward pass of your network, then when combined with the internal clock of the embedded system, you can calculate the approximate time it ...
There are several different ways you can model the state and action spaces in such sequential (extensive-form) environments/games. For environments with small action spaces or those typically introduced to beginning-RL students, the state space and action space remains constant along an agent's trajectory (termed normal form games when there are multiple ...
There is a paper face pose estimation
It uses a very straight forward technique, and very obvious augmentaions to achieve nice results.
You could use exactly the same if you have a tagged dataset for cars rather than for faces.
I was able to reproduce the results myself a while back.
I think the problem can be phrased (more generally) as a Pose Estimation Problem. That term might help in obtaining better search results when searching for relevant papers.
One paper that I found on the given topic was this one. Even if it is maybe (for whatever reason) not what you are looking for precisely, it might still contain valuable references to ...
After my initial comment (where I suggest that it might not be enough info) I believe I actually came up with an idea.
Start with the full set of pokemon. For every possible type, identify the count of pokemon that are strong against that type. For this, you'll end up with a List<(pokemonId, types, List<weakAgainst>)>.
I'm not aware of a direct way for finding the best NN architecture for a given task, but the recommended way, as far as I know, is to devise a network that can overfit the training data, and then apply regularization on top of it.
That way, you can be almost sure you're not underfitting/underperforming due to network capacity.
One way to handle an arbitrarily large sequence is by adding a STOP signal as one possible token in the sequence, just like LSTM.
So you could divide your game in turns:
What you now call a single action (composed by multiple sub-actions) would become a turn.
Now, you can have as many actions you'd like inside a turn.
Each action is simply a list ...
Your best bet would be to formulate the problem in PDDL, which should be fairly easy, and then use a standard planner to generate a plan from that description.
In PDDL you describe the properties and the possible actions, the start state and the goal state, and the planner will then take this to produce a sequence of actions that leads from the start state ...
That's a very nice challenge!
As always, the hardest part is to get a labeled database (maybe by scrapping).
You'd probably need some thousands of drawings and their respective drawer age.
From there, you need to make an image regression model. Here is a simple example that predicts age from a face photo. It's the same principle, but applied to another ...