Artificial Intelligence
Budget: £20 – £250 GBP
I am looking for a freelancer who can develop an Artificial Intelligence (AI) system specifically for Natural Language Processing. The AI will be built using Python as the preferred platform.
• Define the problem domain and datasets) (you are free to choose the problem domain and the dataset that you want to investigate).
• Define questions and analysis tasks (a brief overview of the domain, the questions that are being asked, a list of your objectives and the expected output(s) of your analysis).
• Perform an initial investigation of the dataset and the characteristics of the data. Develop a plan as to how you might transform the data to make it useable.
• Develop a plan as to which artificial intelligence techniques you might use and what sorts of potential observations these can lead to, and how you will evaluate these.
You must use models, these are: perceptron, decision trees, linear regression, support vector machines, random forest, k-nearest neighbour, naïve Bayes, neural networks as well as unsupervised techniques k-means and GMM, and principal component analysis. Most supervised models have both classification and regression variants. You are encouraged to work with neural networks. An additional technique from outside the taught module content might be applied for comparison purposes, if this is done it should be clearly indicated and well justified.
• Split your dataset (train/validate/test, some datasets come pre-split). If you have a holdout test set then you most likely don't want to use this until the near the end of your work
• Perform the analysis. Get the data ready for analysis, carry out your analysis/modelling as needed, validate your results and communicate observations, iterating through this process. Analytical operations can include data processing to an extent that is needed (not all datasets are messy) to prepare a useful and robust dataset to work within, and data derivation (such as feature engineering).
• You might establish a baseline result first, computing metrics on training and validation sets, analyse errors, work on succeeding iterations, and alternative models. (If initial baseline results are amazing and there are no errors is the problem too easy?)
• Generally, be close to your data (visualise the dataset, collect summary statistics, look at errors, analyse how different parameters affect performance, try out different model variants).
Key requirements for the project include:
- Strong knowledge and experience in Natural Language Processing
- Proficiency in Python programming
- Ability to develop a simple chatbot using AI technology
If you have the skills and experience in these areas, please submit your proposal for consideration.
• Define the problem domain and datasets) (you are free to choose the problem domain and the dataset that you want to investigate).
• Define questions and analysis tasks (a brief overview of the domain, the questions that are being asked, a list of your objectives and the expected output(s) of your analysis).
• Perform an initial investigation of the dataset and the characteristics of the data. Develop a plan as to how you might transform the data to make it useable.
• Develop a plan as to which artificial intelligence techniques you might use and what sorts of potential observations these can lead to, and how you will evaluate these.
You must use models, these are: perceptron, decision trees, linear regression, support vector machines, random forest, k-nearest neighbour, naïve Bayes, neural networks as well as unsupervised techniques k-means and GMM, and principal component analysis. Most supervised models have both classification and regression variants. You are encouraged to work with neural networks. An additional technique from outside the taught module content might be applied for comparison purposes, if this is done it should be clearly indicated and well justified.
• Split your dataset (train/validate/test, some datasets come pre-split). If you have a holdout test set then you most likely don't want to use this until the near the end of your work
• Perform the analysis. Get the data ready for analysis, carry out your analysis/modelling as needed, validate your results and communicate observations, iterating through this process. Analytical operations can include data processing to an extent that is needed (not all datasets are messy) to prepare a useful and robust dataset to work within, and data derivation (such as feature engineering).
• You might establish a baseline result first, computing metrics on training and validation sets, analyse errors, work on succeeding iterations, and alternative models. (If initial baseline results are amazing and there are no errors is the problem too easy?)
• Generally, be close to your data (visualise the dataset, collect summary statistics, look at errors, analyse how different parameters affect performance, try out different model variants).
Key requirements for the project include:
- Strong knowledge and experience in Natural Language Processing
- Proficiency in Python programming
- Ability to develop a simple chatbot using AI technology
If you have the skills and experience in these areas, please submit your proposal for consideration.