With the rapid development of large language models (LLMs), traditional full-parameter fine-tuning methods have become increasingly expensive in terms of computational resources and time costs. For this reason, parameter efficient fine-tuning (PEFT) methods have emerged. Among them, Low-Rank Adaptation (LoRA) is one of the current popular PEFT methods, which is widely used in large language models. However, the low-rank update mechanism of LoRA somewhat limits its ability to approximate full-parameter fine-tuning during the training process. In this paper, we propose a novel PEFT framework, MoKA (Mixture of Kronecker Product Adaptation), which combines the Kronecker product with the Mixture-of-Experts (MoE) method. By replacing the low-rank decomposition of the weight update matrix with Kronecker products and utilizing a sparse MoE architecture, MoKA achieves parameter efficiency and better model performance. Additionally, we design an efficient routing module to further compress the parameter size. We conduct extensive experiments on the GLUE benchmark, E2E NLG Challenge, and instruction tuning tasks for LLMs. The results demonstrate that MoKA outperforms existing PEFT methods.

Recently, we have witnessed a significant performance boosting for dialogue response selection task achieved by Cross-Encoder based models. However, such models directly feed the concatenation of context and response into the pre-trained model for interactive inference, ignoring the comprehensively independent representation modeling of context and response. Moreover, randomly sampling negative responses from other dialogue contexts is simplistic, and the learned models have poor generalization capability in realistic scenarios. In this paper, we propose a response selection model called BERT-BC that combines the representation-based Bi-Encoder and interaction-based Cross-Encoder. Three contrastive learning methods are devised for the Bi-Encoder to align context and response to obtain the better semantic representation. Meanwhile, according to the alignment difficulty of context and response semantics, the harder samples are dynamically selected from the same batch with negligible cost and sent to Cross-Encoder to enhance the model’s interactive reasoning ability. Experimental results show that BERT-BC can achieve state-of-the-art performance on three benchmark datasets for multi-turn response selection.

Sentiment analysis has always been an important research direction in natural language processing. The research can be divided into explicit sentiment analysis and implicit sentiment analysis according to whether there are sentiment words in language expression. There have been many research results in explicit sentiment analysis. However, implicit sentiment analysis is rarely studied. Compared with explicit sentiment expression, implicit sentiment expression usually omits a lot of knowledge and common sense, and context also has an important impact on implicit sentiment expression. In this paper, we use a knowledge graph to supplement implicit sentiment expression and propose a novel Implicit Sentiment Analysis model combining Knowledge enhancement and Context features (dubbed KC-ISA). The KC-ISA model can effectively integrate external knowledge and contextual features by the coattention mechanism. Finally, we conduct experiments on the SMP2019 implicit sentiment analysis dataset. Moreover, to verify the generality of the model, we also conduct experiments on two common sentiment analysis datasets. The results on three datasets show that our proposed KC-ISA model can achieve better results on text sentiment analysis.

In this work, we present a solution to real-time monocular dense mapping. A tightly-coupled visual-inertial localization module is designed to provide metric and high-accuracy odometry. A motion stereo algorithm is proposed to take the video input from one camera to produce local depth measurements with semi-global regularization. The local measurements are then integrated into a global map for noise filtering and map refinement. The global map obtained is able to support navigation and obstacle avoidance for aerial robots through our indoor and outdoor experimental verification. Our system runs at 10Hz on an Nvidia Jetson TX1 by properly distributing computation to CPU and GPU. Through onboard experiments, we demonstrate its ability to close the perception-action loop for autonomous aerial robots. We release our implementation as open-source software1.

There have been increasing demands for developing microaerial vehicles with vision-based autonomy for search and rescue missions in complex environments. In particular, the monocular visual-inertial system (VINS), which consists of only an inertial measurement unit (IMU) and a camera, forms a great lightweight sensor suite due to its low weight and small footprint. In this paper, we address two challenges for rapid deployment of monocular VINS: 1) the initialization problem and 2) the calibration problem. We propose a methodology that is able to initialize velocity, gravity, visual scale, and camera-IMU extrinsic calibration on the fly. Our approach operates in natural environments and does not use any artificial markers. It also does not require any prior knowledge about the mechanical configuration of the system. It is a significant step toward plug-and-play and highly customizable visual navigation for mobile robots. We show through online experiments that our method leads to accurate calibration of camera-IMU transformation, with errors less than 0.02 m in translation and 1° in rotation. We compare out method with a state-of-the-art marker-based offline calibration method and show superior results. We also demonstrate the performance of the proposed approach in large-scale indoor and outdoor experiments.

We address the important problem of achieving robust and easy-to-deploy visual state estimation for micro aerial vehicles (MAVs) operating in complex environments. We use a sensor suite consisting of multiple cameras and an IMU to maximize perceptual awareness of the surroundings and provide sufficient redundancy against sensor failures. Our approach starts with an online initialization procedure that simultaneously estimates the transformation between each camera and the IMU, as well as the initial velocity and attitude of the platform, without any prior knowledge about the mechanical configuration of the sensor suite. Based on the initial calibrations, a tightly-coupled, optimization-based, generalized multi-camera-inertial fusion method runs onboard the MAV with online camera-IMU calibration refinement and identification of sensor failures. Our approach dynamically configures the system into monocular, stereo, or other multi-camera visual-inertial settings, with their respective perceptual advantages, based on the availability of visual measurements. We show that even under random camera failures, our method can be used for feedback control of the MAVs. We highlight our approach in challenging indoor-outdoor navigation tasks with large variations in vehicle height and speed, scene depth, and illumination.