Combining Transfer Learning with In-context Learning using Blackbox LLMs for Zero-shot Knowledge Base Question Answering

We address the zero-shot transfer learning setting for the knowledge base question answering (KBQA) problem, where a large volume of labeled training data is available for the source domain, but no such labeled examples are available for the target domain. Transfer learning for KBQA makes use of large volumes of unlabeled data in the target in addition to the labeled data in the source. More recently, few-shot in-context learning using Black-box Large Language Models (BLLMs) has been adapted for KBQA without considering any source domain data. In this work, we show how to meaningfully combine these two paradigms for KBQA so that their benefits add up. Specifically, we preserve the two stage retrieve-then-generate pipeline of supervised KBQA and introduce interaction between in-context learning using BLLMs and transfer learning from the source for both stages. In addition, we propose execution-guided self-refinement using BLLMs, decoupled from the transfer setting. With the help of experiments using benchmark datasets GrailQA as the source and WebQSP as the target, we show that the proposed combination brings significant improvements to both stages and also outperforms by a large margin state-of-the-art supervised KBQA models trained on the source. We also show that in the in-domain setting, the proposed BLLM augmentation significantly outperforms state-of-the-art supervised models, when the volume of labeled data is limited, and also outperforms these marginally even when using the entire large training dataset.

Neuroadaptation in Physical Human-Robot Collaboration

Robots for physical Human-Robot Collaboration (pHRC) systems need to change their behavior and how they operate in consideration of several factors, such as the performance and intention of a human co-worker and the capabilities of different human-co-workers in collision avoidance and singularity of the robot operation. As the system's admittance becomes variable throughout the workspace, a potential solution is to tune the interaction forces and control the parameters based on the operator's requirements. To overcome this issue, we have demonstrated a novel closed-loop-neuroadaptive framework for pHRC. We have applied cognitive conflict information in a closed-loop manner, with the help of reinforcement learning, to adapt to robot strategy and compare this with open-loop settings. The experiment results show that the closed-loop-based neuroadaptive framework successfully reduces the level of cognitive conflict during pHRC, consequently increasing the smoothness and intuitiveness of human-robot collaboration. These results suggest the feasibility of a neuroadaptive approach for future pHRC control systems through electroencephalogram (EEG) signals.

Do I have the Knowledge to Answer? Investigating Answerability of Knowledge Base Questions

When answering natural language questions over knowledge bases (KBs), incompleteness in the KB can naturally lead to many questions being unanswerable. While answerability has been explored in other QA settings, it has not been studied for QA over knowledge bases (KBQA). We first identify various forms of KB incompleteness that can result in a question being unanswerable. We then propose GrailQAbility, a new benchmark dataset, which systematically modifies GrailQA (a popular KBQA dataset) to represent all these incompleteness issues. Testing two state-of-the-art KBQA models (trained on original GrailQA as well as our GrailQAbility), we find that both models struggle to detect unanswerable questions, or sometimes detect them for the wrong reasons. Consequently, both models suffer significant loss in performance, underscoring the need for further research in making KBQA systems robust to unanswerability.

Ransomware Detection using Process Memory

Ransomware attacks have increased significantly in recent years, causing great destruction and damage to critical systems and business operations. Attackers are unfailingly finding innovative ways to bypass detection mechanisms, whichencouraged the adoption of artificial intelligence. However, most research summarizes the general features of AI and induces many false positives, as the behavior of ransomware constantly differs to bypass detection. Focusing on the key indicating features of ransomware becomes vital as this guides the investigator to the inner workings and main function of ransomware itself. By utilizing access privileges in process memory, the main function of the ransomware can be detected more easily and accurately. Furthermore, new signatures and fingerprints of ransomware families can be identified to classify novel ransomware attacks correctly. The current research used the process memory access privileges of the different memory regions of the behavior of an executable to quickly determine its intent before serious harm can occur. To achieve this aim, several well-known machine learning algorithms were explored with an accuracy range of 81.38 to 96.28 percents. The study thus confirms the feasibility of utilizing process memory as a detection mechanism for ransomware.

RoboFly: An insect-sized robot with simplified fabrication that is capable of flight, ground, and water surface locomotion

Aerial robots the size of a honeybee (~100 mg) have advantages over larger robots because of their small size, low mass and low materials cost. Previous iterations have demonstrated controlled flight but were difficult to fabricate because they consisted of many separate parts assembled together. They also were unable to perform locomotion modes besides flight. This paper presents a new design of a 74 mg flapping-wing robot that dramatically reduces the number of parts and simplifies fabrication. It also has a lower center of mass, which allows the robot to additionally land without the need for long legs, even in case of unstable flight. Furthermore, we show that the new design allows for wing-driven ground and air-water interfacial locomotion, improving the versatility of the robot. Forward thrust is generated by increasing the speed of downstroke relative to the upstroke of the flapping wings. This also allows for steering. The ability to land and subsequently move along the ground allows the robot to negotiate extremely confined spaces, underneath obstacles, and to precise locations. We describe the new design in detail and present results demonstrating these capabilities, as well as hovering flight and controlled landing.