Search

The fairness & bias of narrow coverage of AI becomes another challenge for AI researchers. If a commercial AI trains with a biased dataset, there will be severe gender or racial fairness and bias issues. Since the researchers use primary language datasets to train AI, the broad audience cannot be satisfied if a novel LLM (Large Language Model) AI shows a knowledge or creativity limitation on their specific spoken language. Narrow coverage of the LLMs can lead the audience to misinterpretation and confusion if the service involves STT (Speech-To-Text). In this paper, to overcome this issue of data diversity, we propose the idea that the embedded, extracted features have captured semantic proximity information that can be useful to mitigate diversity issues. This project focused on the Korean language food dataset for STT services, where a narrow-trained A.I. is prone to show its limitations, such as lifestyle-related elements. To present our proof of concept, we trained a baseline model, GPT2, with the Korean Wikipedia dataset in 2022. Then, we employed DistilBERT and KoBERT for comparison. The extracted hidden_state_output features from each model were utilized to build feature-extraction-based text search engines. We used the same idea of Local Sensitive Hashing (LSH) but effectively located a similar hash by applying transposed weights. We also present conventional classification benchmarks for performance comparison using top-k measurements, times for training and memory & disc consumptions. In the discussion, we proposed that our idea can mitigate the diversity problem without re-training the model and tokenizer.

This paper presents a dynamic-model-aided navigation (DMAN) method for a small multirotor unmanned aerial vehicle. The method can be used for temporary navigation in cases where location and velocity measurements from external sources, e.g., global navigation satellite system, are missing or unreliable. The method combines proprioceptive measurements with a Kalman filter through a dynamic model to obtain the velocity and location of the vehicle. Acceleration and angular rate measurements from an inertial measurement unit, altitude measurements from a barometric altimeter, and proprioceptive measurements of the revolution speed of propellers are considered in the method. The dynamic model of the aerial vehicle relates the linear and angular velocities of the vehicle with the revolution speed of the propeller. The revolution speed is first converted into a thrust force and torque and then included in the model. The model avoids the singularity problem and describes processes and measurements in a three-dimensional space by representing attitude using quaternions instead of Euler angles. This study details two implementations of the DMAN method: extended Kalman filter (EKF) and unscented Kalman filter (UKF). The dynamic model is incorporated into the process model and measurement model of the implementations. A model that converts the revolution speed of propellers to thrust force and torque has been derived from unmanned aerial vehicle flight experiments. Experiments that implement the proposed method for quadrotor navigation verify the performance and state the limitations of the DMAN method. Compared with previous methods, the proposed method extends the application of DMAN to the three-dimensional space and obtains location and velocity measurements in a world coordinate system.