# AD-Prediction Convolutional Neural Networks for Alzheimer's Disease Prediction Using Brain MRI Image ## Abstract Alzheimers disease (AD) is characterized by severe memory loss and cognitive impairment. It associates with significant brain structure changes, which can be measured by magnetic resonance imaging (MRI) scan. The observable preclinical structure changes provides an opportunity for AD early detection using image classification tools, like convolutional neural network (CNN). However, currently most AD related studies were limited by sample size. Finding an efficient way to train image classifier on limited data is critical. In our project, we explored different transfer-learning methods based on CNN for AD prediction brain structure MRI image. We find that both pretrained 2D AlexNet with 2D-representation method and simple neural network with pretrained 3D autoencoder improved the prediction performance comparing to a deep CNN trained from scratch. The pretrained 2D AlexNet performed even better (**86%**) than the 3D CNN with autoencoder (**77%**). ## Method #### 1. Data In this project, we used public brain MRI data from **Alzheimers Disease Neuroimaging Initiative (ADNI)** Study. ADNI is an ongoing, multicenter cohort study, started from 2004. It focuses on understanding the diagnostic and predictive value of Alzheimers disease specific biomarkers. The ADNI study has three phases: ADNI1, ADNI-GO, and ADNI2. Both ADNI1 and ADNI2 recruited new AD patients and normal control as research participants. Our data included a total of 686 structure MRI scans from both ADNI1 and ADNI2 phases, with 310 AD cases and 376 normal controls. We randomly derived the total sample into training dataset (n = 519), validation dataset (n = 100), and testing dataset (n = 67). #### 2. Image preprocessing Image preprocessing were conducted using Statistical Parametric Mapping (SPM) software, version 12. The original MRI scans were first skull-stripped and segmented using segmentation algorithm based on 6-tissue probability mapping and then normalized to the International Consortium for Brain Mapping template of European brains using affine registration. Other configuration includes: bias, noise, and global intensity normalization. The standard preprocessing process output 3D image files with an uniform size of 121x145x121. Skull-stripping and normalization ensured the comparability between images by transforming the original brain image into a standard image space, so that same brain substructures can be aligned at same image coordinates for different participants. Diluted or enhanced intensity was used to compensate the structure changes. the In our project, we used both whole brain (including both grey matter and white matter) and grey matter only. #### 3. AlexNet and Transfer Learning Convolutional Neural Networks (CNN) are very similar to ordinary Neural Networks. A CNN consists of an input and an output layer, as well as multiple hidden layers. The hidden layers are either convolutional, pooling or fully connected. ConvNet architectures make the explicit assumption that the inputs are images, which allows us to encode certain properties into the architecture. These then make the forward function more efficient to implement and vastly reduce the amount of parameters in the network. #### 3.1. AlexNet The net contains eight layers with weights; the first five are convolutional and the remaining three are fully connected. The overall architecture is shown in Figure 1. The output of the last fully-connected layer is fed to a 1000-way softmax which produces a distribution over the 1000 class labels. AlexNet maximizes the multinomial logistic regression objective, which is equivalent to maximizing the average across training cases of the log-probability of the correct label under the prediction distribution. The kernels of the second, fourth, and fifth convolutional layers are connected only to those kernel maps in the previous layer which reside on the same GPU (as shown in Figure1). The kernels of the third convolutional layer are connected to all kernel maps in the second layer. The neurons in the fully connected layers are connected to all neurons in the previous layer. Response-normalization layers follow the first and second convolutional layers. Max-pooling layers follow both response-normalization layers as well as the fifth convolutional layer. The ReLU non-linearity is applied to the output of every convolutional and fully-connected layer.  The first convolutional layer filters the 224x224x3 input image with 96 kernels of size 11x11x3 with a stride of 4 pixels (this is the distance between the receptive field centers of neighboring neurons in a kernel map). The second convolutional layer takes as input the (response-normalized and pooled) output of the first convolutional layer and filters it with 256 kernels of size 5x5x48. The third, fourth, and fifth convolutional layers are connected to one another without any intervening pooling or normalization layers. The third convolutional layer has 384 kernels of size 3x3x256 connected to the (normalized, pooled) outputs of the second convolutional layer. The fourth convolutional layer has 384 kernels of size 3x3x192 , and the fifth convolutional layer has 256 kernels of size 3x3x192. The fully-connected layers have 4096 neurons each. #### 3.2. Transfer Learning Training an entire Convolutional Network from scratch (with random initialization) is impractical[14] because it is relatively rare to have a dataset of sufficient size. An alternative is to pretrain a Conv-Net on a very large dataset (e.g. ImageNet), and then use the ConvNet either as an initialization or a fixed feature extractor for the task of interest. Typically, there are three major transfer learning scenarios: **ConvNet as fixed feature extractor:** We can take a ConvNet pretrained on ImageNet, and remove the last fully-connected layer, then treat the rest structure as a fixed feature extractor for the target dataset. In AlexNet, this would be a 4096-D vector. Usually, we call these features as CNN codes. Once we get these features, we can train a linear classifier (e.g. linear SVM or Softmax classifier) for our target dataset. **Fine-tuning the ConvNet:** Another idea is not only replace the last fully-connected layer in the classifier, but to also fine-tune the parameters of the pretrained network. Due to overfitting concerns, we can only fine-tune some higher-level part of the network. This suggestion is motivated by the observation that earlier features in a ConvNet contains more generic features (e.g. edge detectors or color blob detectors) that can be useful for many kind of tasks. But the later layer of the network becomes progressively more specific to the details of the classes contained in the original dataset. **Pretrained models:** The released pretrained model is usually the final ConvNet checkpoint. So it is common to see people use the network for fine-tuning. #### 4. 3D Autoencoder and Convolutional Neural Network We take a two-stage approach where we first train a 3D sparse autoencoder to learn filters for convolution operations, and then build a convolutional neural network whose first layer uses the filters learned with the autoencoder.  #### 4.1. Sparse Autoencoder An autoencoder is a 3-layer neural network that is used to extract features from an input such as an image. Sparse representations can provide a simple interpretation of the input data in terms of a small number of \parts by extracting the structure hidden in the data. The autoencoder has an input layer, a hidden layer and an output layer, and the input and output layers have same number of units, while the hidden layer contains more units for a sparse and overcomplete representation. The encoder function maps input x to representation h, and the decoder function maps the representation h to the output x. In our problem, we extract 3D patches from scans as the input to the network. The decoder function aims to reconstruct the input form the hidden representation h. #### 4.2. 3D Convolutional Neural Network Training the 3D convolutional neural network(CNN) is the second stage. The CNN we use in this project has one convolutional layer, one pooling layer, two linear layers, and finally a log softmax layer. After training the sparse autoencoder, we take the weights and biases of the encoder from trained model, and use them a 3D filter of a 3D convolutional layer of the 1-layer convolutional neural network. Figure 2 shows the architecture of the network. #### 5. Tools In this project, we used Nibabel for MRI image processing and PyTorch Neural Networks implementation.

Installer for DataKitchen's Open Source Data Observability Products. Data breaks. Servers break. Your toolchain breaks. Ensure your team is the first to know and the first to solve with visibility across and down your data estate. Save time with simple, fast data quality test generation and execution. Trust your data, tools, and systems end to end.

gRPSeek is a load testing and metrics visualization tool. By utilizing this package, developers can increase observability of their gRPC servers and gain valuable insights into the performance of the services

A curated, opinionated collection of tools and resources dedicated to Performance Engineering, covering both Observability and Performance Testing.

OpenTestability is an open-source tool for structural analysis of digital circuits, enabling computation of SCOAP metrics, Controllibility Observability Probability (COP), reconvergent path detection, and test point insertion.

The "Cloudflare for AI Agents". 7-layer security interceptor, real-time observability dashboard, and automated reliability testing for MCP and AI tool chains. Prevent hallucinations, prompt injection, and destructive tool calls.

Test tool for converting PCAP files and live captures to STIXv2 observables.

# Pandas Homework - Pandas, Pandas, Pandas ## Background The data dive continues! Now, it's time to take what you've learned about Python Pandas and apply it to new situations. For this assignment, you'll need to complete **one of two** (not both) Data Challenges. Once again, which challenge you take on is your choice. Just be sure to give it your all -- as the skills you hone will become powerful tools in your data analytics tool belt. ### Before You Begin 1. Create a new repository for this project called `pandas-challenge`. **Do not add this homework to an existing repository**. 2. Clone the new repository to your computer. 3. Inside your local git repository, create a directory for the Pandas Challenge you choose. Use folder names corresponding to the challenges: **HeroesOfPymoli** or **PyCitySchools**. 4. Add your Jupyter notebook to this folder. This will be the main script to run for analysis. 5. Push the above changes to GitHub or GitLab. ## Option 1: Heroes of Pymoli  Congratulations! After a lot of hard work in the data munging mines, you've landed a job as Lead Analyst for an independent gaming company. You've been assigned the task of analyzing the data for their most recent fantasy game Heroes of Pymoli. Like many others in its genre, the game is free-to-play, but players are encouraged to purchase optional items that enhance their playing experience. As a first task, the company would like you to generate a report that breaks down the game's purchasing data into meaningful insights. Your final report should include each of the following: ### Player Count * Total Number of Players ### Purchasing Analysis (Total) * Number of Unique Items * Average Purchase Price * Total Number of Purchases * Total Revenue ### Gender Demographics * Percentage and Count of Male Players * Percentage and Count of Female Players * Percentage and Count of Other / Non-Disclosed ### Purchasing Analysis (Gender) * The below each broken by gender * Purchase Count * Average Purchase Price * Total Purchase Value * Average Purchase Total per Person by Gender ### Age Demographics * The below each broken into bins of 4 years (i.e. <10, 10-14, 15-19, etc.) * Purchase Count * Average Purchase Price * Total Purchase Value * Average Purchase Total per Person by Age Group ### Top Spenders * Identify the the top 5 spenders in the game by total purchase value, then list (in a table): * SN * Purchase Count * Average Purchase Price * Total Purchase Value ### Most Popular Items * Identify the 5 most popular items by purchase count, then list (in a table): * Item ID * Item Name * Purchase Count * Item Price * Total Purchase Value ### Most Profitable Items * Identify the 5 most profitable items by total purchase value, then list (in a table): * Item ID * Item Name * Purchase Count * Item Price * Total Purchase Value As final considerations: * You must use the Pandas Library and the Jupyter Notebook. * You must submit a link to your Jupyter Notebook with the viewable Data Frames. * You must include a written description of three observable trends based on the data. * See [Example Solution](HeroesOfPymoli/HeroesOfPymoli_starter.ipynb) for a reference on expected format. ## Option 2: PyCitySchools  Well done! Having spent years analyzing financial records for big banks, you've finally scratched your idealistic itch and joined the education sector. In your latest role, you've become the Chief Data Scientist for your city's school district. In this capacity, you'll be helping the school board and mayor make strategic decisions regarding future school budgets and priorities. As a first task, you've been asked to analyze the district-wide standardized test results. You'll be given access to every student's math and reading scores, as well as various information on the schools they attend. Your responsibility is to aggregate the data to and showcase obvious trends in school performance. Your final report should include each of the following: ### District Summary * Create a high level snapshot (in table form) of the district's key metrics, including: * Total Schools * Total Students * Total Budget * Average Math Score * Average Reading Score * % Passing Math (The percentage of students that passed math.) * % Passing Reading (The percentage of students that passed reading.) * % Overall Passing (The percentage of students that passed math **and** reading.) ### School Summary * Create an overview table that summarizes key metrics about each school, including: * School Name * School Type * Total Students * Total School Budget * Per Student Budget * Average Math Score * Average Reading Score * % Passing Math (The percentage of students that passed math.) * % Passing Reading (The percentage of students that passed reading.) * % Overall Passing (The percentage of students that passed math **and** reading.) ### Top Performing Schools (By % Overall Passing) * Create a table that highlights the top 5 performing schools based on % Overall Passing. Include: * School Name * School Type * Total Students * Total School Budget * Per Student Budget * Average Math Score * Average Reading Score * % Passing Math (The percentage of students that passed math.) * % Passing Reading (The percentage of students that passed reading.) * % Overall Passing (The percentage of students that passed math **and** reading.) ### Bottom Performing Schools (By % Overall Passing) * Create a table that highlights the bottom 5 performing schools based on % Overall Passing. Include all of the same metrics as above. ### Math Scores by Grade\*\* * Create a table that lists the average Math Score for students of each grade level (9th, 10th, 11th, 12th) at each school. ### Reading Scores by Grade * Create a table that lists the average Reading Score for students of each grade level (9th, 10th, 11th, 12th) at each school. ### Scores by School Spending * Create a table that breaks down school performances based on average Spending Ranges (Per Student). Use 4 reasonable bins to group school spending. Include in the table each of the following: * Average Math Score * Average Reading Score * % Passing Math (The percentage of students that passed math.) * % Passing Reading (The percentage of students that passed reading.) * % Overall Passing (The percentage of students that passed math **and** reading.) ### Scores by School Size * Repeat the above breakdown, but this time group schools based on a reasonable approximation of school size (Small, Medium, Large). ### Scores by School Type * Repeat the above breakdown, but this time group schools based on school type (Charter vs. District). As final considerations: * Use the pandas library and Jupyter Notebook. * You must submit a link to your Jupyter Notebook with the viewable Data Frames. * You must include a written description of at least two observable trends based on the data. * See [Example Solution](PyCitySchools/PyCitySchools_starter.ipynb) for a reference on the expected format. ## Hints and Considerations * These are challenging activities for a number of reasons. For one, these activities will require you to analyze thousands of records. Hacking through the data to look for obvious trends in Excel is just not a feasible option. The size of the data may seem daunting, but pandas will allow you to efficiently parse through it. * Second, these activities will also challenge you by requiring you to learn on your feet. Don't fool yourself into thinking: "I need to study pandas more closely before diving in." Get the basic gist of the library and then _immediately_ get to work. When facing a daunting task, it's easy to think: "I'm just not ready to tackle it yet." But that's the surest way to never succeed. Learning to program requires one to constantly tinker, experiment, and learn on the fly. You are doing exactly the _right_ thing, if you find yourself constantly practicing Google-Fu and diving into documentation. There is just no way (or reason) to try and memorize it all. Online references are available for you to use when you need them. So use them! * Take each of these tasks one at a time. Begin your work, answering the basic questions: "How do I import the data?" "How do I convert the data into a DataFrame?" "How do I build the first table?" Don't get intimidated by the number of asks. Many of them are repetitive in nature with just a few tweaks. Be persistent and creative! * Expect these exercises to take time! Don't get discouraged if you find yourself spending hours initially with little progress. Force yourself to deal with the discomfort of not knowing and forge ahead. Consider these hours an investment in your future! * As always, feel encouraged to work in groups and get help from your TAs and Instructor. Just remember, true success comes from mastery and _not_ a completed homework assignment. So challenge yourself to truly succeed! ### Copyright Trilogy Education Services © 2019. All Rights Reserved.

A repo to test out various observability tools.

making a simple l4 proxy mode load balancer to practise testing and observability tools

Part I - WeatherPy In this example, you’ll be creating a Python script to visualize the weather of 500+ cities across the world of varying distance from the equator. To accomplish this, you’ll be utilizing a simple Python library, the OpenWeatherMap API, and a little common sense to create a representative model of weather across world cities. Your first objective is to build a series of scatter plots to showcase the following relationships: Temperature (F) vs. Latitude Humidity (%) vs. Latitude Cloudiness (%) vs. Latitude Wind Speed (mph) vs. Latitude After each plot add a sentence or too explaining what the code is and analyzing. Your next objective is to run linear regression on each relationship, only this time separating them into Northern Hemisphere (greater than or equal to 0 degrees latitude) and Southern Hemisphere (less than 0 degrees latitude): Northern Hemisphere - Temperature (F) vs. Latitude Southern Hemisphere - Temperature (F) vs. Latitude Northern Hemisphere - Humidity (%) vs. Latitude Southern Hemisphere - Humidity (%) vs. Latitude Northern Hemisphere - Cloudiness (%) vs. Latitude Southern Hemisphere - Cloudiness (%) vs. Latitude Northern Hemisphere - Wind Speed (mph) vs. Latitude Southern Hemisphere - Wind Speed (mph) vs. Latitude After each pair of plots explain what the linear regression is modelling such as any relationships you notice and any other analysis you may have. Your final notebook must: Randomly select at least 500 unique (non-repeat) cities based on latitude and longitude. Perform a weather check on each of the cities using a series of successive API calls. Include a print log of each city as it’s being processed with the city number and city name. Save a CSV of all retrieved data and a PNG image for each scatter plot. Part II - VacationPy Now let’s use your skills in working with weather data to plan future vacations. Use jupyter-gmaps and the Google Places API for this part of the assignment. Note: if you having trouble displaying the maps try running jupyter nbextension enable --py gmaps in your environment and retry. Create a heat map that displays the humidity for every city from the part I of the homework. heatmap Narrow down the DataFrame to find your ideal weather condition. For example: A max temperature lower than 80 degrees but higher than 70. Wind speed less than 10 mph. Zero cloudiness. Drop any rows that don’t contain all three conditions. You want to be sure the weather is ideal. Note: Feel free to adjust to your specifications but be sure to limit the number of rows returned by your API requests to a reasonable number. Using Google Places API to find the first hotel for each city located within 5000 meters of your coordinates. Plot the hotels on top of the humidity heatmap with each pin containing the Hotel Name, City, and Country. hotel map As final considerations: Create a new GitHub repository for this project called API-Challenge (note the kebab-case). Do not add to an existing repo You must complete your analysis using a Jupyter notebook. You must use the Matplotlib or Pandas plotting libraries. For Part I, you must include a written description of three observable trends based on the data. You must use proper labeling of your plots, including aspects like: Plot Titles (with date of analysis) and Axes Labels. For max intensity in the heat map, try setting it to the highest humidity found in the data set. Hints and Considerations The city data you generate is based on random coordinates as well as different query times; as such, your outputs will not be an exact match to the provided starter notebook. You may want to start this assignment by refreshing yourself on the geographic coordinate system. Next, spend the requisite time necessary to study the OpenWeatherMap API. Based on your initial study, you should be able to answer basic questions about the API: Where do you request the API key? Which Weather API in particular will you need? What URL endpoints does it expect? What JSON structure does it respond with? Before you write a line of code, you should be aiming to have a crystal clear understanding of your intended outcome. A starter code for Citipy has been provided. However, if you’re craving an extra challenge, push yourself to learn how it works: citipy Python library. Before you try to incorporate the library into your analysis, start by creating simple test cases outside your main script to confirm that you are using it correctly. Too often, when introduced to a new library, students get bogged down by the most minor of errors – spending hours investigating their entire code – when, in fact, a simple and focused test would have shown their basic utilization of the library was wrong from the start. Don’t let this be you! Part of our expectation in this challenge is that you will use critical thinking skills to understand how and why we’re recommending the tools we are. What is Citipy for? Why would you use it in conjunction with the OpenWeatherMap API? How would you do so? In building your script, pay attention to the cities you are using in your query pool. Are you getting coverage of the full gamut of latitudes and longitudes? Or are you simply choosing 500 cities concentrated in one region of the world? Even if you were a geographic genius, simply rattling 500 cities based on your human selection would create a biased dataset. Be thinking of how you should counter this. (Hint: Consider the full range of latitudes). Once you have computed the linear regression for one chart, the process will be similar for all others. As a bonus, try to create a function that will create these charts based on different parameters. Remember that each coordinate will trigger a separate call to the Google API. If you’re creating your own criteria to plan your vacation, try to reduce the results in your DataFrame to 10 or fewer cities. Lastly, remember – this is a challenging activity. Push yourself! If you complete this task, then you can safely say that you’ve gained a strong mastery of the core foundations of data analytics and it will only go better from here. Good luck!

STOTS (Structured Threat Observation Tool Set) is a collection of tools that allow users in a test environment to create STIX v2 observable objects.

A fast, async Python HTTP benchmarking tool inspired by wrk and Apache ab. Supports concurrent connections, latency breakdown, percentile stats, SLO thresholds, rate limiting, scripted scenarios, live TUI dashboard, multi-URL testing, distributed master/worker mode, and observability export (OpenTelemetry, Prometheus).

Locust Telemetry is a plugin for the Locust load-testing framework that emits structured telemetry and metrics. It supports JSON log-based telemetry and OpenTelemetry metrics, enabling easy integration with any observability tool or dashboard, including Grafana, Prometheus, Datadog, and New Relic.

A tool for serving mock metrics and emitting mock logs for testing observability things.

A tool to determine how extreme transformations propagate to observable points. Reneri consolidates Decartes with suggestions to improve the test suite

## WeatherPy In this example, you'll be creating a Python script to visualize the weather of 500+ cities across the world of varying distance from the equator. To accomplish this, you'll be utilizing a [simple Python library](https://pypi.python.org/pypi/citipy), the [OpenWeatherMap API](https://openweathermap.org/api), and a little common sense to create a representative model of weather across world cities. Your objective is to build a series of scatter plots to showcase the following relationships: * Temperature (F) vs. Latitude * Humidity (%) vs. Latitude * Cloudiness (%) vs. Latitude * Wind Speed (mph) vs. Latitude Your final notebook must: * Randomly select **at least** 500 unique (non-repeat) cities based on latitude and longitude. * Perform a weather check on each of the cities using a series of successive API calls. * Include a print log of each city as it's being processed with the city number and city name. * Save both a CSV of all data retrieved and png images for each scatter plot. As final considerations: * You must complete your analysis using a Jupyter notebook. * You must use the Matplotlib or Pandas plotting libraries. * You must include a written description of three observable trends based on the data. * You must use proper labeling of your plots, including aspects like: Plot Titles (with date of analysis) and Axes Labels. * See [Example Solution](WeatherPy_Example.pdf) for a reference on expected format. ## Hints and Considerations * You may want to start this assignment by refreshing yourself on the [geographic coordinate system](http://desktop.arcgis.com/en/arcmap/10.3/guide-books/map-projections/about-geographic-coordinate-systems.htm). * Next, spend the requisite time necessary to study the OpenWeatherMap API. Based on your initial study, you should be able to answer basic questions about the API: Where do you request the API key? Which Weather API in particular will you need? What URL endpoints does it expect? What JSON structure does it respond with? Before you write a line of code, you should be aiming to have a crystal clear understanding of your intended outcome. * A starter code for Citipy has been provided. However, if you're craving an extra challenge, push yourself to learn how it works: [citipy Python library](https://pypi.python.org/pypi/citipy). Before you try to incorporate the library into your analysis, start by creating simple test cases outside your main script to confirm that you are using it correctly. Too often, when introduced to a new library, students get bogged down by the most minor of errors -- spending hours investigating their entire code -- when, in fact, a simple and focused test would have shown their basic utilization of the library was wrong from the start. Don't let this be you! * Part of our expectation in this challenge is that you will use critical thinking skills to understand how and why we're recommending the tools we are. What is Citipy for? Why would you use it in conjunction with the OpenWeatherMap API? How would you do so? * In building your script, pay att

🛠️ A Python tool for generating bulk logs & metrics in Google Cloud. Simulate historical & live data to test alerts, refine LQL/SQL/PromQL queries, troubleshoot observability setups, and create structured data for training & automation.

Testing Tool for LLM Observability

open source load testing + LLM observability tool

CLI tool for deploying production-ready APISIX infrastructure on Hetzner Cloud with automatic scaling, observability, and load testing.

(Rust) Coastguard is a monitoring and observability tool. Log events, detect errors and write rigorous tests

A CLI tool that performs Copilot-assisted audits of any codebase across 6 key dimensions: dead code, architectural leaks, selector fragility, routing inconsistencies, missing tests, and observability gaps.

A framework for chatbot testing and evaluation using Arize Phoenix. This project leverages Arize's observability tools to analyze chatbot performance, with integrated reference data stored in a local database for validation and comparison.

Build a scalable, testable, and observable data stack from ingestion to analytics using industry-standard tools and architectural patterns.

Hybrid code audit system powered by deterministic MCP tools and conditional hotspot exploration. Audits 12 domains: security, tests, architecture, conventions, performance, documentation, tech-debt, accessibility, API design, observability, i18n, and UX consistency.

Your AI Language Tutor for foreign languages. This microservices app showcases modern SWE practices (polyglot monorepo, Docker, automated testing, security, observability). Developed via a gamified "Polyglot Dev Quest" with AI assistance, it's an effective language tool and a strong developer portfolio piece.

This repository is a structured learning hub for agentic AI systems and GenAI engineering. It includes notes, playground experiments, end-to-end projects, and production patterns covering LangChain, LangGraph, RAG, MCP tools, observability, testing/evals, and cloud integrations (Azure AI, AWS Bedrock, Databricks, vector DBs).

BrowserStack Test Observability for WebdriverIO - a next-generation tool which brings together test reporting, precision debugging, flaky test detection and more... all on the same easy-to-use dashboard.

Test automation, observability and tooling.