Conference Paper

Feature-Adaptive and Data-Scalable In-Context Learning

January 2024

DOI:10.18653/v1/2024.acl-long.81

Conference: Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)

Authors:

Show all 5 authorsHide

A Survey on In-context Learning

Conference Paper

Jan 2024

ResearchGate has not been able to resolve any references for this publication.

Article

Full-text available

A Feature-Adaptive and Scalable Hardware Trojan Detection Framework For Third-party IPs Utilizing Mu...

December 2024 · Journal of Electronic Testing

The trustworthiness of integrated circuits is susceptible to the hardware Trojans in the third-party intellectual property (IP) cores. Various hardware Trojan detection approaches have been proposed to identify the Trojans from a given netlist at the design stage, however, such methods suffer from low detection accuracy, poor scalability, small-scale circuit, and inability to cover more types of ... [Show full abstract] Trojans. To address these issues, a hardware Trojan detection framework for third-party IPs is proposed, which is capable of detecting the various types of Trojans with high accuracy and scalable to find the unknown Trojans. Some typical gate-level Trojans in the benchmark circuits are analyzed and 13 common features are introduced as the Trojan features. Furthermore, a feature extraction method is presented to extract 13 features of each node from 23 gate-level circuits and form the Trojan’s feature set. Besides, a feature rebalancing method is proposed to balance the distribution of genuine and Trojan class, and a feature combination selection method is proposed to choose the optimal Trojan feature combination of each Trojan circuit, and thus the Trojan’s feature database is established. Finally, the Trojan detection is formulated as the anomalous feature identification problem, and a feature-adaptive classifier based on the random forest is proposed to identify various types and features of Trojans. Simulation results of 23 Trojan circuits demonstrate that the proposed approach achieves a comparable precision (TPR=90.48%), recall (TNR=97.92%), and accuracy (ACC=97.96%) compared to the existing methods. Besides, the proposed approach is not only able to detect the existing Trojans in the circuits that scale from 0.2K to 120K nodes, but more importantly, it also has the unique advantage of identifying more types of Trojans by adding new Trojan features to the Trojan feature database.

View full-text

Article

Full-text available

Kernel Context Recommender System (KCR): A Scalable Context-Aware Recommender System Algorithm

February 2019 · IEEE Access

Recommender systems are intelligent data mining applications that deal with the issue of information overload significantly. The available literature discusses several methodologies to generate recommendations and proposes different techniques in accordance to users’ needs. The majority of the work in the recommender system domain focuses on increasing the recommendation accuracy by employing ... [Show full abstract] several proposed approaches where the main motive remains to maximize the accuracy of recommendations while ignoring other design objectives, such as, a user’s and item’s context. The biggest challenge for a recommender system is to produce meaningful recommendations by using contextual user-item rating information. A context is a vast term that may consider various aspects; for example, a user’s social circle, time, mood, location, weather, company, day type, an item’s genre, location and language. Typically, the rating behavior of users varies under different contexts. From this line of research, we have proposed a new algorithm, namely KCR, which is a flexible, fast, and accurate kernel mapping framework that recognizes the importance of context and incorporates the contextual information using kernel trick while making predictions. We have benchmarked our proposed algorithm with pre- and post-filtering approaches as they have been the favorite approaches in the literature to solve the context-aware recommendation problem. Our experiments reveal that considering the contextual information can increase the performance of a system and provide better, relevant, and meaningful results on various evaluation metrics.

View full-text

Article

Full-text available

Feature-adaptive FPN with multiscale context integration for underwater object detection

September 2024 · Earth Science Informatics

Underwater object detection is vital for diverse applications, from studies in marine biology to underwater robotics. However, underwater environments pose unique challenges, including reduced visibility due to color distortion, light attenuation, and complex backgrounds. Traditional computer vision methods have limitations, prompting the implementation of deep learning, for underwater object ... [Show full abstract] detection. Despite progress, challenges persist, such as visual degradation, scale variations, diverse marine species, and complex backgrounds. To address these issues, we propose Feature-Adaptive FPN with Multiscale Context Integration (FA-FPN-MCI), a novel deep-learning algorithm aimed at enhancing both detection and domain generalization performance. We integrate the Style Normalization and Restitution (SNR) module for domain generalization, Receptive Field Blocks (RFBs) for fine-grained detail capture, and a twin-branch Global Context Module (TBGCM) for multiscale context information. We enhance lateral connections within the Feature Pyramid Network (FPN) with deformable convolution. Experimental outcome reveal that the proposed method attains mean average precision of 84.2%. Additionally, other performance metrics were evaluated, and outperforming all other methods used for comparison.

View full-text

Article

Full-text available

Scalable context-based motion vector coding for video compression

January 1970

State of the art video compression algorithms use motion compensation. Using such method gives better results by in-creasing the temporal correlation between pixels from con-secutive frames. Most of the effort in increasing the perfor-mance of existing algorithms concentrates on the improve-ment of motion vector uses and residual coding. In this paper we propose a motion vectors computation and ... [Show full abstract] coding scheme for video compression which combines con-strained block matching and scalable coding based on a three-step search-like decomposition.

View full-text

Feature-Adaptive and Data-Scalable In-Context Learning

No full-text available

Recommended publications

A Feature-Adaptive and Scalable Hardware Trojan Detection Framework For Third-party IPs Utilizing Mu...

Kernel Context Recommender System (KCR): A Scalable Context-Aware Recommender System Algorithm

Feature-adaptive FPN with multiscale context integration for underwater object detection

Scalable context-based motion vector coding for video compression