Deep learning is one of the variety of machinelearning in which the higher level features from the data are extracted through numerous layers of processing. The idea of artificial intelligence (AI) was first proposed...
详细信息
An IDS is a system that helps in detecting any kind of doubtful activity on a computer network. It is capable of identifying suspicious activities at both the levels i.e. locally at the system level and in transit at ...
详细信息
作者：
Zhang, QiWang, YifeiWang, YisenPeking Univ
Sch Intelligence Sci & Technol Natl Key Lab Gen Artificial Intelligence Beijing Peoples R China Peking Univ
Sch Math Sci Beijing Peoples R China Peking Univ
Inst Artificial Intelligence Beijing Peoples R China
Multi-modal contrastive learning (MMCL) has recently garnered considerable interest due to its superior performance in visual tasks, achieved by embedding multi-modal data, such as visual-language pairs. However, ther...
详细信息
Multi-modal contrastive learning (MMCL) has recently garnered considerable interest due to its superior performance in visual tasks, achieved by embedding multi-modal data, such as visual-language pairs. However, there still lack theoretical understandings of how MMCL extracts useful visual representation from multi-modal pairs, and particularly, how MMCL outperforms previous approaches like self-supervised contrastive learning (SSCL). In this paper, by drawing an intrinsic connection between MMCL and asymmetric matrix factorization, we establish the first generalization guarantees of MMCL for visual downstream tasks. Based on this framework, we further unify MMCL and SSCL by showing that MMCL implicitly performs SSCL with (pseudo) positive pairs induced by text pairs. Through this unified perspective, we characterize the advantage of MMCL by showing that text pairs induce more semantically consistent and diverse positive pairs, which, according to our analysis, provably benefit downstream generalization. Inspired by this finding, we propose several methods to significantly improve the downstream performance of SSCL on ImageNet by leveraging multi-modal information. Code is available at https://***/PKU-ML/CLIP-Help-SimCLR.
Power electronics is used everywhere be it to charge our smartphone, electric vehicle, power system etc., thus making condition monitoring of power electronics crucial for reliability of the system. Considering the wi...
详细信息
Heart disease is considered as one of the common health problem, and machinelearning can be a powerful tool for reducing the burden of disease. Heart Disease Prediction Model using machinelearning is a process of us...
详细信息
Mineral prospectivity mapping (MPM) has been an essential part of mineral exploration, various algorithms have been introduced for detecting mineralization related anomalies from multi-geoinformation including geology...
详细信息
Due to the influence of many high random factors on the new energy power generation system, the electric energy output by the generator is extremely unstable, which increases the difficulty of predicting the power gen...
详细信息
In mass manufacturing of jewelry, the gross loss is estimated before manufacturing to calculate the wax weight of the pattern that would be investment casted to make multiple identical pieces of jewelry. machine learn...
详细信息
Artificial Intelligence (AI) built-in Consumer Electronics is popular, but it is hard to test and evaluate AI-based system with the existing performance metrics. Even though AI-based systems are implemented in softwar...
详细信息
Inter-processes shared information between applications that are running over multiple platforms over the Internet is the most appropriate environment to hack these data. Users emit a huge amount of confidential data ...
详细信息
暂无评论