8
votes
Accepted
Why is the merged neural network of AlphaGo Zero more efficient than two separate neural networks?
Why has this merge proven beneficial?
If you think about the shared Value/Policy network as consisting of a shared component (the Residual Network layers) with a Value and Policy component on top ...
- 426
4
votes
Accepted
Transposition table is only used for roughly 17% of the nodes - is this expected?
I don't think that's necessarily a strange number. It's impossible for anyone to really tell you whether that 17% is "correct" or not without reproducing it, which would require much more info (...
- 10.4k
3
votes
Accepted
Is it possible to guide a reinforcement learning algorithm?
The programmer already guides the RL algorithm (or agent) by specifying the reward function. However, the reward function alone may not be sufficient to learn efficiently and fast, as you correctly ...
- 41.1k
3
votes
Why don't people always use TensorFlow Lite, if it doesn't decrease the accuracy of the models?
This partly answer to question 1.
There is no general rule concerning accuracy or size of the model. It depends on the training data and the processed data.
The lightest is your model compared to the ...
- 241
2
votes
Can neural networks efficiently solve the traveling salesmen problem?
There is some recent work addressing this issue, to learn the conditional probability of an output sequence with elements that are discrete tokens corresponding to positions in an input sequence. See ...
- 364
2
votes
Accepted
Can neural networks efficiently solve the traveling salesmen problem?
To the best of my knowledge, there isn't any difference between the algorithmic methods and the NN methods. Those that can solve in polynomial time do not give a precise solution. Those that do give a ...
- 271
2
votes
Accepted
Which are more memory efficient: uninformed or informed search algorithms?
Uninformed Search Techniques
Breadth-First Search needs to store a frontier of nodes to visit next (where "visit" basically means: see if it's the goal, generate its children and add to ...
- 10.4k
2
votes
Accepted
In the multi-head attention mechanism of the transformer, why do we need both $W_i^Q$ and ${W_i^K}^T$?
I'll use notation from the paper you cited, and any other readers should refer to the paper (widely available) for definitions of notation. The utility of using $W^Q$ and $W^K$, rather than $W$, lies ...
- 148
2
votes
Is it possible to guide a reinforcement learning algorithm?
Here are two very related interesting papers:
Learning from Human Preferences
Improving Reinforcement Learning with Human Input
- 461
1
vote
Accepted
Matrix Multiplication using a Neural Network
AlphaTensor by DeepMind is likely the most notable attempt to speed up matrix multiplication using AI.
As a note on the second question, likely the advancement here would be to attempt to analyze what ...
- 1,477
1
vote
Compare the efficiency of a trained ML model with a non-learning-based method for solving the same problem
The most generic answer to this question is:
the same metrics you use to evaluate the quality of your model during training or in test phase. (Plus the timing of inference if you're referring to ...
- 5,378
1
vote
Accepted
A comparison of Expert Systems and Machine Learning approaches in terms of run-time-efficiency and time/space complexity
My feeling from what I have generally read about machine learning approaches is that these adapt better to more complex problems with higher dimensionalities.
Your intuition seems to align with the ...
- 348
1
vote
Using AI to enhance customer service
AI can transform the customer experience is by providing personalized content. For example, When you see video recommendation on YouTube, you'll know that it's from AI technology. I recommend you to ...
- 11
1
vote
Using AI to enhance customer service
This sounds to me like a use case for a chatbot. You would have different intents reflecting the types of user queries that your system can respond to. The intent matching can be done by pattern ...
- 5,397
1
vote
Accepted
Why don't people always use TensorFlow Lite, if it doesn't decrease the accuracy of the models?
I have explored the AI for edge devices. My findings for tflite model.
TFLite is just the tool suite to convert ...
- 826
1
vote
Accepted
Why is informed search more efficient than uninformed search?
There are several informed and uninformed search algorithms. They do not all have the same time and space complexity (which also depends on the specific implementation). I could come up with an ...
- 41.1k
1
vote
Are there any load balancing approaches for employee cases based on complexity score?
It sounds like you're looking at the Partition Problem. https://en.wikipedia.org/wiki/Partition_problem
The task of slicing one set into N sets so that each set is equal or as close to equal as ...
- 146
1
vote
Accepted
how fast does need to be an AI agent to be efficient?
My understanding of efficiency in this context is in regards to optimization of algorithms as opposed to hardware speed, which is more of a "brute force" component. GPUs may be more energy efficient, ...
- 6,235
Only top scored, non community-wiki answers of a minimum length are eligible