41 skills found · Page 1 of 2
fjb040911 / AI Rulesai-rules is a governance framework designed to solve "Architectural Decay" in AI-driven development. It forces AI Agents (Cursor, Windsurf, Copilot) to respect your project's boundaries, UI libraries, and design patterns.
akirarika / Milkio🌟 A TypeScript Framework - Now, break the boundaries between Frontend and Backend
Flafla2 / Vive TeleporterA framework for Unity3D that automatically generates teleporter boundaries and facilitates the teleporter mechanic, similarly to Valve's "The Lab"
evstack / Ev NodeA modular framework for building performant networks, pushing the boundary of today in order to build the apps of tomorrow.
molyswu / Hand Detectionusing Neural Networks (SSD) on Tensorflow. This repo documents steps and scripts used to train a hand detector using Tensorflow (Object Detection API). As with any DNN based task, the most expensive (and riskiest) part of the process has to do with finding or creating the right (annotated) dataset. I was interested mainly in detecting hands on a table (egocentric view point). I experimented first with the [Oxford Hands Dataset](http://www.robots.ox.ac.uk/~vgg/data/hands/) (the results were not good). I then tried the [Egohands Dataset](http://vision.soic.indiana.edu/projects/egohands/) which was a much better fit to my requirements. The goal of this repo/post is to demonstrate how neural networks can be applied to the (hard) problem of tracking hands (egocentric and other views). Better still, provide code that can be adapted to other uses cases. If you use this tutorial or models in your research or project, please cite [this](#citing-this-tutorial). Here is the detector in action. <img src="images/hand1.gif" width="33.3%"><img src="images/hand2.gif" width="33.3%"><img src="images/hand3.gif" width="33.3%"> Realtime detection on video stream from a webcam . <img src="images/chess1.gif" width="33.3%"><img src="images/chess2.gif" width="33.3%"><img src="images/chess3.gif" width="33.3%"> Detection on a Youtube video. Both examples above were run on a macbook pro **CPU** (i7, 2.5GHz, 16GB). Some fps numbers are: | FPS | Image Size | Device| Comments| | ------------- | ------------- | ------------- | ------------- | | 21 | 320 * 240 | Macbook pro (i7, 2.5GHz, 16GB) | Run without visualizing results| | 16 | 320 * 240 | Macbook pro (i7, 2.5GHz, 16GB) | Run while visualizing results (image above) | | 11 | 640 * 480 | Macbook pro (i7, 2.5GHz, 16GB) | Run while visualizing results (image above) | > Note: The code in this repo is written and tested with Tensorflow `1.4.0-rc0`. Using a different version may result in [some errors](https://github.com/tensorflow/models/issues/1581). You may need to [generate your own frozen model](https://pythonprogramming.net/testing-custom-object-detector-tensorflow-object-detection-api-tutorial/?completed=/training-custom-objects-tensorflow-object-detection-api-tutorial/) graph using the [model checkpoints](model-checkpoint) in the repo to fit your TF version. **Content of this document** - Motivation - Why Track/Detect hands with Neural Networks - Data preparation and network training in Tensorflow (Dataset, Import, Training) - Training the hand detection Model - Using the Detector to Detect/Track hands - Thoughts on Optimizations. > P.S if you are using or have used the models provided here, feel free to reach out on twitter ([@vykthur](https://twitter.com/vykthur)) and share your work! ## Motivation - Why Track/Detect hands with Neural Networks? There are several existing approaches to tracking hands in the computer vision domain. Incidentally, many of these approaches are rule based (e.g extracting background based on texture and boundary features, distinguishing between hands and background using color histograms and HOG classifiers,) making them not very robust. For example, these algorithms might get confused if the background is unusual or in situations where sharp changes in lighting conditions cause sharp changes in skin color or the tracked object becomes occluded.(see [here for a review](https://www.cse.unr.edu/~bebis/handposerev.pdf) paper on hand pose estimation from the HCI perspective) With sufficiently large datasets, neural networks provide opportunity to train models that perform well and address challenges of existing object tracking/detection algorithms - varied/poor lighting, noisy environments, diverse viewpoints and even occlusion. The main drawbacks to usage for real-time tracking/detection is that they can be complex, are relatively slow compared to tracking-only algorithms and it can be quite expensive to assemble a good dataset. But things are changing with advances in fast neural networks. Furthermore, this entire area of work has been made more approachable by deep learning frameworks (such as the tensorflow object detection api) that simplify the process of training a model for custom object detection. More importantly, the advent of fast neural network models like ssd, faster r-cnn, rfcn (see [here](https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/detection_model_zoo.md#coco-trained-models-coco-models) ) etc make neural networks an attractive candidate for real-time detection (and tracking) applications. Hopefully, this repo demonstrates this. > If you are not interested in the process of training the detector, you can skip straight to applying the [pretrained model I provide in detecting hands](#detecting-hands). Training a model is a multi-stage process (assembling dataset, cleaning, splitting into training/test partitions and generating an inference graph). While I lightly touch on the details of these parts, there are a few other tutorials cover training a custom object detector using the tensorflow object detection api in more detail[ see [here](https://pythonprogramming.net/training-custom-objects-tensorflow-object-detection-api-tutorial/) and [here](https://towardsdatascience.com/how-to-train-your-own-object-detector-with-tensorflows-object-detector-api-bec72ecfe1d9) ]. I recommend you walk through those if interested in training a custom object detector from scratch. ## Data preparation and network training in Tensorflow (Dataset, Import, Training) **The Egohands Dataset** The hand detector model is built using data from the [Egohands Dataset](http://vision.soic.indiana.edu/projects/egohands/) dataset. This dataset works well for several reasons. It contains high quality, pixel level annotations (>15000 ground truth labels) where hands are located across 4800 images. All images are captured from an egocentric view (Google glass) across 48 different environments (indoor, outdoor) and activities (playing cards, chess, jenga, solving puzzles etc). <img src="images/egohandstrain.jpg" width="100%"> If you will be using the Egohands dataset, you can cite them as follows: > Bambach, Sven, et al. "Lending a hand: Detecting hands and recognizing activities in complex egocentric interactions." Proceedings of the IEEE International Conference on Computer Vision. 2015. The Egohands dataset (zip file with labelled data) contains 48 folders of locations where video data was collected (100 images per folder). ``` -- LOCATION_X -- frame_1.jpg -- frame_2.jpg ... -- frame_100.jpg -- polygons.mat // contains annotations for all 100 images in current folder -- LOCATION_Y -- frame_1.jpg -- frame_2.jpg ... -- frame_100.jpg -- polygons.mat // contains annotations for all 100 images in current folder ``` **Converting data to Tensorflow Format** Some initial work needs to be done to the Egohands dataset to transform it into the format (`tfrecord`) which Tensorflow needs to train a model. This repo contains `egohands_dataset_clean.py` a script that will help you generate these csv files. - Downloads the egohands datasets - Renames all files to include their directory names to ensure each filename is unique - Splits the dataset into train (80%), test (10%) and eval (10%) folders. - Reads in `polygons.mat` for each folder, generates bounding boxes and visualizes them to ensure correctness (see image above). - Once the script is done running, you should have an images folder containing three folders - train, test and eval. Each of these folders should also contain a csv label document each - `train_labels.csv`, `test_labels.csv` that can be used to generate `tfrecords` Note: While the egohands dataset provides four separate labels for hands (own left, own right, other left, and other right), for my purpose, I am only interested in the general `hand` class and label all training data as `hand`. You can modify the data prep script to generate `tfrecords` that support 4 labels. Next: convert your dataset + csv files to tfrecords. A helpful guide on this can be found [here](https://pythonprogramming.net/creating-tfrecord-files-tensorflow-object-detection-api-tutorial/).For each folder, you should be able to generate `train.record`, `test.record` required in the training process. ## Training the hand detection Model Now that the dataset has been assembled (and your tfrecords), the next task is to train a model based on this. With neural networks, it is possible to use a process called [transfer learning](https://www.tensorflow.org/tutorials/image_retraining) to shorten the amount of time needed to train the entire model. This means we can take an existing model (that has been trained well on a related domain (here image classification) and retrain its final layer(s) to detect hands for us. Sweet!. Given that neural networks sometimes have thousands or millions of parameters that can take weeks or months to train, transfer learning helps shorten training time to possibly hours. Tensorflow does offer a few models (in the tensorflow [model zoo](https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/detection_model_zoo.md#coco-trained-models-coco-models)) and I chose to use the `ssd_mobilenet_v1_coco` model as my start point given it is currently (one of) the fastest models (read the SSD research [paper here](https://arxiv.org/pdf/1512.02325.pdf)). The training process can be done locally on your CPU machine which may take a while or better on a (cloud) GPU machine (which is what I did). For reference, training on my macbook pro (tensorflow compiled from source to take advantage of the mac's cpu architecture) the maximum speed I got was 5 seconds per step as opposed to the ~0.5 seconds per step I got with a GPU. For reference it would take about 12 days to run 200k steps on my mac (i7, 2.5GHz, 16GB) compared to ~5hrs on a GPU. > **Training on your own images**: Please use the [guide provided by Harrison from pythonprogramming](https://pythonprogramming.net/training-custom-objects-tensorflow-object-detection-api-tutorial/) on how to generate tfrecords given your label csv files and your images. The guide also covers how to start the training process if training locally. [see [here] (https://pythonprogramming.net/training-custom-objects-tensorflow-object-detection-api-tutorial/)]. If training in the cloud using a service like GCP, see the [guide here](https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/running_on_cloud.md). As the training process progresses, the expectation is that total loss (errors) gets reduced to its possible minimum (about a value of 1 or thereabout). By observing the tensorboard graphs for total loss(see image below), it should be possible to get an idea of when the training process is complete (total loss does not decrease with further iterations/steps). I ran my training job for 200k steps (took about 5 hours) and stopped at a total Loss (errors) value of 2.575.(In retrospect, I could have stopped the training at about 50k steps and gotten a similar total loss value). With tensorflow, you can also run an evaluation concurrently that assesses your model to see how well it performs on the test data. A commonly used metric for performance is mean average precision (mAP) which is single number used to summarize the area under the precision-recall curve. mAP is a measure of how well the model generates a bounding box that has at least a 50% overlap with the ground truth bounding box in our test dataset. For the hand detector trained here, the mAP value was **0.9686@0.5IOU**. mAP values range from 0-1, the higher the better. <img src="images/accuracy.jpg" width="100%"> Once training is completed, the trained inference graph (`frozen_inference_graph.pb`) is then exported (see the earlier referenced guides for how to do this) and saved in the `hand_inference_graph` folder. Now its time to do some interesting detection. ## Using the Detector to Detect/Track hands If you have not done this yet, please following the guide on installing [Tensorflow and the Tensorflow object detection api](https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/installation.md). This will walk you through setting up the tensorflow framework, cloning the tensorflow github repo and a guide on - Load the `frozen_inference_graph.pb` trained on the hands dataset as well as the corresponding label map. In this repo, this is done in the `utils/detector_utils.py` script by the `load_inference_graph` method. ```python detection_graph = tf.Graph() with detection_graph.as_default(): od_graph_def = tf.GraphDef() with tf.gfile.GFile(PATH_TO_CKPT, 'rb') as fid: serialized_graph = fid.read() od_graph_def.ParseFromString(serialized_graph) tf.import_graph_def(od_graph_def, name='') sess = tf.Session(graph=detection_graph) print("> ====== Hand Inference graph loaded.") ``` - Detect hands. In this repo, this is done in the `utils/detector_utils.py` script by the `detect_objects` method. ```python (boxes, scores, classes, num) = sess.run( [detection_boxes, detection_scores, detection_classes, num_detections], feed_dict={image_tensor: image_np_expanded}) ``` - Visualize detected bounding detection_boxes. In this repo, this is done in the `utils/detector_utils.py` script by the `draw_box_on_image` method. This repo contains two scripts that tie all these steps together. - detect_multi_threaded.py : A threaded implementation for reading camera video input detection and detecting. Takes a set of command line flags to set parameters such as `--display` (visualize detections), image parameters `--width` and `--height`, videe `--source` (0 for camera) etc. - detect_single_threaded.py : Same as above, but single threaded. This script works for video files by setting the video source parameter videe `--source` (path to a video file). ```cmd # load and run detection on video at path "videos/chess.mov" python detect_single_threaded.py --source videos/chess.mov ``` > Update: If you do have errors loading the frozen inference graph in this repo, feel free to generate a new graph that fits your TF version from the model-checkpoint in this repo. Use the [export_inference_graph.py](https://github.com/tensorflow/models/blob/master/research/object_detection/export_inference_graph.py) script provided in the tensorflow object detection api repo. More guidance on this [here](https://pythonprogramming.net/testing-custom-object-detector-tensorflow-object-detection-api-tutorial/?completed=/training-custom-objects-tensorflow-object-detection-api-tutorial/). ## Thoughts on Optimization. A few things that led to noticeable performance increases. - Threading: Turns out that reading images from a webcam is a heavy I/O event and if run on the main application thread can slow down the program. I implemented some good ideas from [Adrian Rosebuck](https://www.pyimagesearch.com/2017/02/06/faster-video-file-fps-with-cv2-videocapture-and-opencv/) on parrallelizing image capture across multiple worker threads. This mostly led to an FPS increase of about 5 points. - For those new to Opencv, images from the `cv2.read()` method return images in [BGR format](https://www.learnopencv.com/why-does-opencv-use-bgr-color-format/). Ensure you convert to RGB before detection (accuracy will be much reduced if you dont). ```python cv2.cvtColor(image_np, cv2.COLOR_BGR2RGB) ``` - Keeping your input image small will increase fps without any significant accuracy drop.(I used about 320 x 240 compared to the 1280 x 720 which my webcam provides). - Model Quantization. Moving from the current 32 bit to 8 bit can achieve up to 4x reduction in memory required to load and store models. One way to further speed up this model is to explore the use of [8-bit fixed point quantization](https://heartbeat.fritz.ai/8-bit-quantization-and-tensorflow-lite-speeding-up-mobile-inference-with-low-precision-a882dfcafbbd). Performance can also be increased by a clever combination of tracking algorithms with the already decent detection and this is something I am still experimenting with. Have ideas for optimizing better, please share! <img src="images/general.jpg" width="100%"> Note: The detector does reflect some limitations associated with the training set. This includes non-egocentric viewpoints, very noisy backgrounds (e.g in a sea of hands) and sometimes skin tone. There is opportunity to improve these with additional data. ## Integrating Multiple DNNs. One way to make things more interesting is to integrate our new knowledge of where "hands" are with other detectors trained to recognize other objects. Unfortunately, while our hand detector can in fact detect hands, it cannot detect other objects (a factor or how it is trained). To create a detector that classifies multiple different objects would mean a long involved process of assembling datasets for each class and a lengthy training process. > Given the above, a potential strategy is to explore structures that allow us **efficiently** interleave output form multiple pretrained models for various object classes and have them detect multiple objects on a single image. An example of this is with my primary use case where I am interested in understanding the position of objects on a table with respect to hands on same table. I am currently doing some work on a threaded application that loads multiple detectors and outputs bounding boxes on a single image. More on this soon.
matthieuaussal / GypsilabThe gypsilab project is an open-source MATLAB toolbox for fast numerical computation with finite element, boundary element and ray-tracing methods. Accessible with a high-level programming language, it gives a useful framework for fast prototyping. Initially designed for numerical acoustics, many physics problems can also be addressed.
QWED-AI / Qwed VerificationAISecOps (AI Security Operations) framework for deterministic verification of AI systems. QWED verifies LLM outputs using math, logic, and symbolic execution — creating an auditable trust boundary for agentic AI systems. Not generation. Verification.
ShadowHackrs / Prompt Generatingis an elite, highly advanced command-line framework designed for generating complex prompt structures. It is built to push the boundaries of AI models by testing their alignment, safety filters, and operational limits. By utilizing state-of-the-art obfuscation techniques, multi-layered theoretical bypass architectures, and linguistic fusion
x-orpheus / Catch React ErrorA framework using React Boundary handles error easily
hetu-project / Hetu ChaoschainThe Agentic chaoschain is an innovative framework powered by AI Agents that includes governance, consensus, proposals, and dispute resolution. It allows Agents to help humans explore and expand the boundaries of social contracts.
danielcamposramos / Knowledge3DWeb knowledge is fragmented — duplicated across fonts, embeddings, metadata, and renderings. Humans see pixels, AI sees tokens, neither shares the source. Knowledge3D: a sovereign GPU-native reference implementation for W3C PM-KR, where humans and AI consume the same procedural knowledge from one source.
Go-Quant / GoquantGoQuant is a powerful Go framework designed for financial data analysis and visualizations, with no boundaries!
llmir / MultitaskOCTAThis repository is an official PyTorch implementation of the paper "BSDA-Net: A Boundary Shape and Distance Aware Joint Learning Framework for Segmenting and Classifying OCTA Images", MICCAI 2021.
earlence-security / CellmateCellmate is a sandboxing framework for BUAs that enforces strict boundaries on their behavior, ensuring safety even in the worst-case execution scenario.
TS-CUBED / HaemoFoamHaemodynamics simulation framework based on OpenFOAM. Includes particle migration model and advanced haemorheology models, as well as windkessel boundary conditions.
ai-agents-simplified / Awesome AI AgentsA curated list of AI agents, frameworks, and tools that automate tasks, enhance workflows, and push the boundaries of artificial intelligence ⚙️
Genaios / TextMachinaA modular and extensible Python framework, designed to aid in the creation of high-quality, unbiased datasets to build robust models for MGT-related tasks such as detection, attribution, and boundary detection.
Nikkitaseth / ProjectAlphaPYTHON CODE WALKTHROUGH Data Sourcing In order to run a discounted cash flow model (DCF), I needed data, so I found a free API that provided us with everything I needed. I wrote a code that saved every financial statement of every company in a separate text file. In this code, I asked to ping the API’s URL for every ticker, open a text file for one of the financial statements for one company ticker, dump all the data found by the code into this file, and close it. This process was repeated for every company in our company list and every statement I have a code for. By doing so I Ire able to store the data for every company locally and did not need to ping the API every time I ran our code. Once all the financial data for each company was stored in form of a balance sheet, income statement, cash flow statement, and company profile text file, I needed to pick out specific items required for our DCF model. Thus, I defined the functions that selected all required items from the respective financial statements of each company and assigned them to a variable using utils.py. Discounted Cash Flow Model First of all, I needed to import the functions I defined in utils.py before defining the DCF model function, which would run for every company in our list. Next, I ensured to have 5 consecutive years of past data to compute the average. Thus, the first few lines of code checked whether the last year on record was 2019 from which point I would go back 5 years; if the last year was 2018, this would be taken as the first data entry from which I would go back 5 years. The second part mentioned above is important because companies file their 10-K, i.e. their annual report, at different times throughout the year so there may be companies that already filed their reports while others had not. After this step, five-year averages of every item’s percentage of revenue Ire calculated as Ill as the average revenue growth over the same period. These items included EBIT, depreciation & amortization, capital expenditures, and the change in net working capital. Once that was done, there Ire only three variables missing before calculating free cash flows for the next few years: a discount or hurdle rate; industry-specific perpetual growth rates; and a tax rate. After these three variables Ire set up, the next step was to calculate the free cash flows to the firm (fcff) for the next 5 years and determine the terminal value at the end of the period using the growth rate for the corresponding industry. For the former, I use a loop to calculate the fcff for all the year, discount it, and add it to one variable called fcffpv. Once the terminal value was calculated, these two additional numbers captured the enterprise value of the firm. Since I Ire interested in the equity value, I subtracted debt and add cash, which left us with the equity value. In one final step, I divided this value by the number of shares to end up with an intrinsic value per share. After calculating the intrinsic value per share, I compared it to the current share price with two additions. First, I added a buffer to minimize our downside risk for inaccuracy in calculations, which is called the margin of safety. Here, the intrinsic value should at least be 115% of the current share price. I also set an upper limit at 130% to ensure I would not include companies with extraordinarily high valuations, compared to their current price. If the share price calculated fell within this window, I added its ticker to a dataframe, which was the last step in the function. As such, the DCF function would run for every company and provide a dataframe with the tickers of all those companies that Ire undervalued at the time and fell within the 115% - 130% range. Portfolio Optimization The dataframe with the tickers of all the undervalued companies that was previously created has now become the portfolio, which I converted into a list and used as the source for further optimization that is about to come. Some general inputs for the rest of the code Ire the start and end date of the data I requested for optimization, as Ill as the risk-free rate and the number of simulations I wanted to run our optimizations for. Now that the general framework has been created, it is time to choose some conditioning variables to measure the performance of investment in one sector or across a combination of some/all sectors, respectively. Project Alpha uses the following conditioning variables to optimize its portfolios: • Sharpe Ratio: It measures the performance of an investment compared to the risk-free asset, i.e. the 10-year Treasury Bond, after adjusting for its risk factor or standard deviation. The Sharpe ratio would be given a higher Iight for investors who have a higher risk tolerance. In terms of code, I used the bt package to retrieve the data betIen the predetermined start and end date for the companies in our ticker list. This data was then used to find the portfolio with the highest Sharpe ratio. For that, random Iights Ire assigned to each company and the ratio was computed. After running the number of simulations previously determined, the Iights with the highest Sharpe ratio will be located using loc() and labeled ‘sharpe_portfolio’ which is a dataframe containing the excess return, the volatility, Sharpe ratio, as Ill as the Iights for every company. I also located the portfolio with the loIst volatility, put it in a dataframe called ‘min_volatility_port’ which has the same attributes. The rest of the code of this segment simply created a picture with all the portfolios generated, displaying the efficient frontier and highlighting the portfolio with the highest Sharpe ratio and loIst volatility. • Value at Risk (VaR): VaR was chosen as a diagnostic tool to assess the model. In our case, it basically indicated the percentage of time in which a loss greater than 1% would occur over a period of 5 years. Its limitation is that although it measures how bad the best of the bad is, it does not measure how bad it can get, meaning the worst of the worst. In regards to the code, I first requested the adjusted closing for the companies in our ticker list in the determined time horizon. I then retrieved the Iights from our Sharpe portfolio, set the number of days I wanted to simulate as Ill as the cutoff, before calculating the returns of every company in every period; here: daily. Thereafter, I created a new variable called ‘sigma’, which was be a copy of our return variable, in order to ensure the right format and type for our Monte Carlo loop. The simulation is pretty straight forward, as it measures how many runs the returns fall within 1% or outside of it. I then Iighed the resulting returns by the Iight of the company in the portfolio and whenever the portfolio return was outside the set boundary, it would count as a ‘bad simulation’. Once that is done, the number of bad simulations was divided by the total number of simulations to end up with a percentage of how many simulations were bad, which equals our VaR • Treynor Ratio: For the investors that already have a perfectly diversified portfolio and would like to add more assets to it, there would be a higher Iight on the Treynor ratio. It basically uses beta as a risk factor because it carries the risk relative to the market, instead of standard deviation as in Sharpe, meaning only systematic or non-diversifiable risk. For the code, I first calculated the portfolio’s beta. For that, I defined a function ‘beta’ that reads the beta of every company and returns it. The next step is to run a loop that would enter the beta of every company in our ticker list into a new dataframe. After setting the index equal to the tickers and transposing the Sharpe portfolio Iights, I can concat the two thus resulting in two columns: one is the beta of every company and the second is the corresponding Iight in the portfolio. I then created a third column as the product of columns one and two. The sum of all entries in that column is the portfolio beta, which was then used as the denominator for the ratio. The nominator was already calculated as ‘Excess Return’ in the Sharpe portfolio. • Sortino Ratio: The Sortino ratio measures only the downside risk (downside deviation or semi-deviation) by measuring returns against a minimum acceptable return, 𝜏. It is surprising to know that most of the industry ignores the total number of periods taken and just calculates the downside deviation by choosing the periods with downside risk, which results in misleading results. Project Alpha uses all the periods to calculate the same, so as to have an advantage over those robo-advisors/financial advisors that do not follow this process. The alpha in the future would be generated by going long on companies with high correct Sortino and low incorrect Sortino as they are undervalued, and shorting those with low correct Sortino and high incorrect Sortino as these are overvalued. The Sortino ratio would be given more Iight for investors who are more risk averse. This part of the code started with retrieving the data for our benchmark, the S&P 500, for the period and the calculating the average daily and annual return. After that, I calculate the portfolio returns, ‘returns[“Returns”]’, by adding the products of every company’s Iight times its return, which gave us the portfolio return for every period. From here, I calculated the downside risk by comparing the portfolio return in every period to the daily average return of our benchmark in a for loop. Before I did that, I defined a new variable called ‘semi’, which is a data series and will be filled with whatever comes out of the loop every single time. If the portfolio return minus the average daily return of the benchmark was greater than 0 – meaning the portfolio earned more than the average of the S&P500 – the value for the period was set to 0 and added to the semi data series. If it is 0, which is extremely unlikely, but whatever, it would also be 0. If it is less than 0, hoIver, which indicates underperformance, I would square the portfolio return, which already gives us the semi variance I need for our next step. From here, I can simply take the square root of the average of the ‘semi’ data series to get the daily downside risk and multiplying it by the square root of 252, which gives us the annual number. After that, I have all the numbers to calculate the Sortino ratio. • Information Ratio: The information ratio measures the portfolio returns compared to the returns of a benchmark index, i.e. S&P500, after adjusting for its additional risk. It only looks at the excess return of the portfolio over the benchmark and the volatility or risk associated with it. I already have all the inputs I need to calculate his ratio. Thus, I simply created a new dataframe with the portfolio returns of every period and the benchmark returns of every period. To find the excess return, i.e. the nominator, I simply subtracted the latter from the former and assigned it to a new variable, which I called ‘excess_return’. The nominator would be the average return of the portfolio minus the average return of the benchmark, and the denominator would be the standard deviation of the ‘excess_return’ series. Finally, I printed short sentences with the results for every conditioning variable just described as an output in the console.
ihamburglar / Dotnetover.nethttps://jimshaver.net/2018/02/22/net-over-net-breaking-the-boundaries-of-the-net-framework/
PKU-SEC-Lab / ODB DLLMImplementation of "Orchestrating Dual-Boundaries: An Arithmetic Intensity Inspired Acceleration Framework for Diffusion Language Models"
