Motivated by the recent empirical successes of deep generative models, we study the computational complexity of the following unsupervised learning problem. For an unknown neural network $F:\mathbb{R}^d\to\mathbb{R}^{d'}$, let $D$ be the distribution over $\mathbb{R}^{d'}$ given by pushing the standard Gaussian $\mathcal{N}(0,\textrm{Id}_d)$ through $F$. Given i.i.d. samples from $D$, the goal is to output any distribution close to $D$ in statistical distance. We show under the statistical query (SQ) model that no polynomial-time algorithm can solve this problem even when the output coordinates of $F$ are one-hidden-layer ReLU networks with $\log(d)$ neurons. Previously, the best lower bounds for this problem simply followed from lower bounds for supervised learning and required at least two hidden layers and $\mathrm{poly}(d)$ neurons [Daniely-Vardi '21, Chen-Gollakota-Klivans-Meka '22]. The key ingredient in our proof is an ODE-based construction of a compactly supported, piecewise-linear function $f$ with polynomially-bounded slopes such that the pushforward of $\mathcal{N}(0,1)$ under $f$ matches all low-degree moments of $\mathcal{N}(0,1)$.
Neural Simpletrons - Minimalistic Directed Generative Networks for Learning with Few Labels
Carl von Ossietzky University of Oldenburg(奥尔登堡卡尔·冯·奥西茨基大学);Frankfurt Institute for Advanced Studies (FIAS)(法兰克福高等研究所(FIAS));Technical University of Berlin(柏林工业大学)
Physics-informed generative AI for semiconductor manufacturing: Enforcing hard physical constraints in generative models by construction
School of Electrical and Computer Engineering, University of Oklahoma(俄克拉荷马大学电气与计算机工程学院);Center for Quantum Research and Technology, University of Oklahoma(俄克拉荷马大学量子研究与技术中心);Intelligent Neuromorphic and Quantum Understanding for Innovative Research and Engineering (INQUIRE) Laboratory(创新研究与工程智能神经形态与量子理解实验室);Material Science and Engineering Program, University of Oklahoma, Norman, OK 73019 USA(俄克拉荷马大学材料科学与工程项目,Norman, OK 73019 USA)
When Hard Negatives Hurt: Bridging the Generative-Discriminative Gap in Hard Negative Synthesis for Retrieval
Shenzhen International Graduate School, Tsinghua University(清华大学深圳国际研究生院);Huawei Technologies Co., Ltd.(华为技术有限公司);City University of Hong Kong(香港城市大学);School of Cyber Science and Technology, Sun Yat-sen University(中山大学信息科学与技术学院);School of Intelligence Science and Technology, Nanjing University(南京大学智能科学与技术学院);The Hong Kong University of Science and Technology(香港科学与技术大学);Huawei Noah’s Ark Lab(华为诺亚实验室)
cMelGAN: An Efficient Conditional Generative Model Based on Mel Spectrograms
University of Toronto(多伦多大学)
在机器学习领域分析音乐是一个非常困难的问题,需要考虑许多限制。音频数据的性质,其非常高的维度和广泛变化的结构尺度,是它如此难以建模的主要原因之一。机器学习在音乐中有许多应用,例如对一段音乐的情绪进行分类,有条件的音乐生成或流行预测。该项目的目标是开发一个基于Mel声谱图的音乐体裁条件生成模型,并通过将其与使用基于音符表示的现有生成音乐模型进行比较来评估其性能。我们最初实现了一个自回归的、基于RNN的生成模型,称为MelNet。然而,由于其速度慢且输出保真度低,我们决定创建一种新的全卷积架构,该架构基于Mel GAN [4]和条件GAN架构,称为cMel GAN。
GANDSE: Generative Adversarial Network based Design Space Exploration for Neural Network Accelerator Design
Nanjing University(南京大学);Zhejiang University(浙江大学);Key Laboratory of Collaborative Sensing and Autonomous Unmanned Systems of Zhejiang Province(浙江省协同感知与自主无人系统重点实验室)
Accelerating Generative Neural Networks on Unmodified Deep Learning Processors -- A Software Approach
Hefei University of Technology(合肥工业大学);Institute of Computing Technology, Chinese Academy of Sciences(中国科学院计算技术研究所);National University of Singapore(新加坡国立大学)
Style-based quantum generative adversarial networks for Monte Carlo events
Technology Innovation Institute(技术创新研究院);Universitat de Barcelona(巴塞罗那大学);CERN(欧洲核子研究组织);National Yang Ming Chiao Tung University(阳明交通大学);Università degli Studi di Milano(米兰大学);INFN Sezione di Milano(意大利国家核物理研究所米兰分部)