no code implementations • 28 Feb 2024 • Mengnan Zhao, Lihe Zhang, Yuqiu Kong, BaoCai Yin
To tackle this issue, we initially employ the feature activation differences between clean and adversarial examples to analyze the underlying causes of CO. Intriguingly, our findings reveal that CO can be attributed to the feature coverage induced by a few specific pathways.
1 code implementation • 3 Feb 2024 • Mengnan Zhao, Lihe Zhang, Tianhang Zheng, Yuqiu Kong, BaoCai Yin
Large-scale diffusion models, known for their impressive image generation capabilities, have raised concerns among researchers regarding social impacts, such as the imitation of copyrighted artistic styles.
no code implementations • 9 Dec 2023 • Mengnan Zhao, Lihe Zhang, Yuqiu Kong, BaoCai Yin
It enhances the initial instance positions through weighted farthest point sampling and further refines the instance positions and proposals using aggregation averaging and center matching.
1 code implementation • ICCV 2023 • Mengnan Zhao, Lihe Zhang, Yuqiu Kong, BaoCai Yin
To address this, we analyze the training process of prior FAT work and observe that catastrophic overfitting is accompanied by the appearance of loss convergence outliers.
1 code implementation • 17 Oct 2021 • Mengnan Zhao, Lihe Zhang, Yuqiu Kong, BaoCai Yin
Specifically, the transient learning network considers transient memories as a static knowledge graph, and the time-aware recurrent evolution network learns representations through a sequence of recurrent evolution units from long-short-term memories.
2 code implementations • 6 Mar 2021 • Shabnam Daghaghi, Nicholas Meisburger, Mengnan Zhao, Yong Wu, Sameh Gobriel, Charlie Tai, Anshumali Shrivastava
Our work highlights several novel perspectives and opportunities for implementing randomized algorithms for deep learning on modern CPUs.
no code implementations • 1 Jan 2021 • Shabnam Daghaghi, Tharun Medini, Beidi Chen, Mengnan Zhao, Anshumali Shrivastava
Softmax classifiers with a very large number of classes naturally occur in many applications such as natural language processing and information retrieval.
no code implementations • 31 Dec 2020 • Shabnam Daghaghi, Tharun Medini, Nicholas Meisburger, Beidi Chen, Mengnan Zhao, Anshumali Shrivastava
Unfortunately, due to the dynamically updated parameters and data samples, there is no sampling scheme that is provably adaptive and samples the negative classes efficiently.
no code implementations • 12 Oct 2019 • Mustafa D. Kaba, Mengnan Zhao, Rene Vidal, Daniel P. Robinson, Enrique Mallada
In the case of the partial discrete Fourier transform, our characterization of the largest sparsity pattern that can be recovered requires the unknown signal to be real and its dimension to be a prime number.
no code implementations • WS 2018 • Mengnan Zhao, Aaron J. Masino, Christopher C. Yang
We investigate the quality of task specific word embeddings created with relatively small, targeted corpora.