Search for: All records

Zero shot time series forecasting is the challenge of forecasting future values of a time dependent sequence without having access to any historical data from the target series during model training. This setting differs from the traditional domain of time series forecasting, where models are typically trained using large volumes of historical data, from the same distribution. Zero shot time series forecasting models are designed to generalize to unseen time series by leveraging their knowledge learned from other, similar series during training. This work proposes two architectures designed for zero shot time series forecasting: zSiFT and zSHiFT. Both architectures use transformer models arranged in a Siamese network configuration. The zSHiFT architecture differs from the zSiFT by the introduction of a hierarchical transformer component to the Siamese network. These architectures are evaluated on vehicular traffic data in California available from the Caltrans Performance Measurement System (PeMS). The models were trained with traffic flow data collected in one region of California and then are evaluated by forecasting traffic in other regions. Forecast accuracy was evaluated at different time horizons (4 to 48 hours). The zSiFT model achieves a Mean Absolute Error (MAE) that is 8.3% lower than the baseline LSTM with attention mechanism model. The zSiFT model achieves an MAE which is 6.6% lower than zSHiFT’s MAE. 

Personalized cooking recipe recommendation systems offer the potential to improve dietary choices for unhoused individuals and those transitioning out of homelessness. However, existing systems often neglect the needs of users with minimal cooking experience, providing little guidance during meal preparation. This study proposes the development of an intelligent cooking assistant system designed to offer realtime, step-by-step support throughout the cooking process. The system integrates a Raspberry Pi 5 mini-computer with a Raspberry Pi AI HAT+ (AI HAT+) and Raspberry Pi AI Camera (AI Camera), strategically mounted above the cooking area to continuously monitor culinary activity. At its core, the assistant utilizes a deep learning image classification model built on Ultralytics’ You Only Look Once version 11 (YOLO11) framework, trained on a curated dataset of 1,339 images collected during the preparation of chicken teriyaki and pasta dishes. The model achieved 100% precision and 99% recall of identifying all six cooking states utilized in this work, resulting in an average confidence accuracy of 91% during real-time tests. The system is intended to enable greater culinary independence among individuals with little cooking experience, such as those affected by long-term homelessness. 

The rise of generative Artificial Intelligence (AI) has created the possibility of presenting novel recipes, i.e., recipes that do not exactly match any known recipe and this has led to the creation of AI-based recipe recommendation systems. AI-based recipe recommendation has the possibility of accommodating a variety of preferences – including a person’s current health (e.g., diabetes), health goals (e.g., weight loss), taste preferences, cultural or ethical needs (e.g., vegan diet). However, unlike recipes recommended or created by a human dietitian, recipes created by generative AI do not guarantee accuracy, i.e., the generated recipe may not meet the requirements specified by the user. This work quantitatively evaluates how closely recipes generated by OpenAI’s GPT4 large language models, created in response to specific prompts, match known recipes in a collection of human-curated recipes. The prompts also include requests for a health condition, diabetes. The recipes are from the largest online community of home cooks sharing recipes (www.allrecipes.com) and the Mayo Clinic’s collection of diabetes meal plan recipes. Recipes from these sources are assumed to be authoritative and thus are used as ground truth for this evaluation. Quantitative evaluation using NLP techniques (Named Entity Recognition (NER) to extract each ingredient from the recipes and cosine similarity metrics) enable computing the quality of the AI results along a continuum. Our results show that the ingredients list in the AI-generated recipe matches 67-88% with the ingredients in the equivalent recipe in the ground truth database. The corresponding cooking directions match 64-86%. Ingredients in recipes generated by AI for diabetics match those in known recipes in our ground truth datasets at widely varying levels: between 26-83%. The quantitative evaluation is used to inform the development of a web-based personalized recipe recommendation system for diabetics that uses OpenAI’s GPT4 model for recipe generation. 

This research paper describes the design of a pill dispensing device that can assist people with physical or cognitive limitations in taking their prescribed medications. The design is based on the communication between two devices for the purpose of dispensing pills at a scheduled time and identifying if these pills had been properly consumed within a specified time frame. The two devices are based on Arduino RP2040 connect microcontrollers and implement several sensors in the aid of dispensing and detecting of pill consumption. The sensors implemented are an IMU, and distances sensors, such as an ultrasonic sensor and an IR proximity sensor, additionally a real time clock module and stepper motor have been included in the design for the scheduling and dispensing of the pills. The two devices will communicate using Bluetooth for low energy devices (BLE) and the purpose of the devices is to provide aid to the intended target audience in achieving a healthier lifestyle. 

This research paper describes the design of a device that can assist seniors or people with physical or cognitive limitations to take their prescribed medications that are in the form of pills on time while verifying that such pills have been actually consumed. The design consists of a portable smart pill dispenser that will rest on a base, allowing it to dispense pills into a smart cup. The smart pill dispenser uses a stepper motor to rotate to a desired pills based on a specific time slot/day of the week. The smart cup attached to the pill box uses an accelerometer, gyroscope, and an IR proximity sensor to detect if a user is taking the medication by how much the smart cup is lifted and tilted. The smart cup will inform the smart pill dispenser if the pills are properly consumed or not, thus, allowing the device to potentially aid the patients to have a healthier life. 

Few-shot machine learning attempts to predict outputs given only a very small number of training examples. The key idea behind most few-shot learning approaches is to pre-train the model with a large number of instances from a different but related class of data, classes for which a large number of instances are available for training. Few-shot learning has been most successfully demonstrated for classification problems using Siamese deep learning neural networks. Few-shot learning is less extensively applied to time-series forecasting. Few-shot forecasting is the task of predicting future values of a time-series even when only a small set of historic time-series is available. Few-shot forecasting has applications in domains where a long history of data is not available. This work describes deep neural network architectures for few-shot forecasting. All the architectures use a Siamese twin network approach to learn a difference function between pairs of time-series, rather than directly forecasting based on historical data as seen in traditional forecasting models. The networks are built using Long short-term memory units (LSTM). During forecasting, a model is able to forecast time-series types that were never seen in the training data by using the few available instances of the new time-series type as reference inputs. The proposed architectures are evaluated on Vehicular traffic data collected in California from the Caltrans Performance Measurement System (PeMS). The models were trained with traffic flow data collected at specific locations and then are evaluated by predicting traffic at different locations at different time horizons (0 to 12 hours). The Mean Absolute Error (MAE) was used as the evaluation metric and also as the loss function for training. The proposed architectures show lower prediction error than a baseline nearest neighbor forecast model. The prediction error increases at longer time horizons.