arXiv subjects

Fan Wu

Publications and source records attributed to Fan Wu.

At least 289 records · Page 16Linked to original sources

arXiv2023

DC-CCL: Device-Cloud Collaborative Controlled Learning for Large Vision Models

Many large vision models have been deployed on the cloud for real-time services. Meanwhile, fresh samples are continuously generated on the served mobile device. How to leverage the device-side samples to improve the cloud-side large model becomes a practical requirement, but falls into the dilemma of no raw sample up-link and no large model down-link. Specifically, the user may opt out of sharing raw samples with the cloud due to the concern of privacy or communication overhead, while the size of some large vision models far exceeds the mobile device's runtime capacity. In this work, we propose a device-cloud collaborative controlled learning framework, called DC-CCL, enabling a cloud-side large vision model that cannot be directly deployed on the mobile device to still benefit from the device-side local samples. In particular, DC-CCL vertically splits the base model into two submodels, one large submodel for learning from the cloud-side samples and the other small submodel for learning from the device-side samples and performing device-cloud knowledge fusion. Nevertheless, on-device training of the small submodel requires the output of the cloud-side large submodel to compute the desired gradients. DC-CCL thus introduces a light-weight model to mimic the large cloud-side submodel with knowledge distillation, which can be offloaded to the mobile device to control its small submodel's optimization direction. Given the decoupling nature of two submodels in collaborative learning, DC-CCL also allows the cloud to take a pre-trained model and the mobile device to take another model with a different backbone architecture.

cs.LG↗

arXiv2023

An efficient model algorithm for two-dimensional field-effect transistors

Two-dimensional materials-based field-effect transistors (2DM-FETs) exhibit both ambipolar and unipolar transport types. To physically and compactly cover both cases, we put forward a quasi-Fermi-level phase space (QFLPS) approach to model the ambipolar effect in our previous work. This work aims to further improve the QFLPS model's numerical aspect so that the model can be implanted into the standard circuit simulator. We first rigorously derive the integral-free formula for the drain-source current to achieve this goal. It is more friendly to computation than the integral form. Besides, it explicitly gives the correlation terms between the electron and hole components. Secondly, to work out the boundary values required by the new expressions, we develop a fast evaluation algorithm for the surface electrostatic potential based on the zero-temperature limit property of the 2DM-FET system. By calibrating the model with the realistic device data of black phosphorus (BP) and monolayer molybdenum disulfide (ML-MoS2) FETs, the completed algorithm is tested against practical cases. The results show a typical superiority to the benchmark algorithm by two orders of magnitude in time consumption can be achieved while keeping a high accuracy with 7 to 9 significant digits.

cond-mat.mtrl-sci↗

arXiv2023

To Store or Not? Online Data Selection for Federated Learning with Limited Storage

Machine learning models have been deployed in mobile networks to deal with massive data from different layers to enable automated network management and intelligence on devices. To overcome high communication cost and severe privacy concerns of centralized machine learning, federated learning (FL) has been proposed to achieve distributed machine learning among networked devices. While the computation and communication limitation has been widely studied, the impact of on-device storage on the performance of FL is still not explored. Without an effective data selection policy to filter the massive streaming data on devices, classical FL can suffer from much longer model training time ($4\times$) and significant inference accuracy reduction ($7\%$), observed in our experiments. In this work, we take the first step to consider the online data selection for FL with limited on-device storage. We first define a new data valuation metric for data evaluation and selection in FL with theoretical guarantees for speeding up model convergence and enhancing final model accuracy, simultaneously. We further design {\ttfamily ODE}, a framework of \textbf{O}nline \textbf{D}ata s\textbf{E}lection for FL, to coordinate networked devices to store valuable data samples. Experimental results on one industrial dataset and three public datasets show the remarkable advantages of {\ttfamily ODE} over the state-of-the-art approaches. Particularly, on the industrial dataset, {\ttfamily ODE} achieves as high as $2.5\times$ speedup of training time and $6\%$ increase in inference accuracy, and is robust to various factors in practical environments.

cs.LG↗

arXiv2023

Criticality-Enhanced Quantum Sensing in the Anisotropic Quantum Rabi Model

Quantum systems that undergo quantum phase transitions exhibit divergent susceptibility and can be exploited as probes to estimate physical parameters. We generalize the dynamic framework for criticality-enhanced quantum sensing by the quantum Rabi model (QRM) to its anisotropic counterpart and derive the correspondingly analytical expressions for the quantum Fisher information (QFI). We find that the contributions of the rotating-wave and counterrotating-wave interaction terms are symmetric at the limit of the infinite ratio of qubit frequency to field frequency, with the QFI reaching a maximum for the isotropic quantum Rabi model. At finite frequency scaling, we analytically derive the inverted variance of higher-order correction and find that it is more affected by the rotating-wave coupling than by the counterrotating-wave coupling.

quant-ph↗

arXiv2023

Learning Optimal Deterministic Auctions with Correlated Valuation Distributions

In mechanism design, it is challenging to design the optimal auction with correlated values in general settings. Although value distribution can be further exploited to improve revenue, the complex correlation structure makes it hard to acquire in practice. Data-driven auction mechanisms, powered by machine learning, enable to design auctions directly from historical auction data, without relying on specific value distributions. In this work, we design a learning-based auction, which can encode the correlation of values into the rank score of each bidder, and further adjust the ranking rule to approach the optimal revenue. We strictly guarantee the property of strategy-proofness by encoding game theoretical conditions into the neural network structure. Furthermore, all operations in the designed auctions are differentiable to enable an end-to-end training paradigm. Experimental results demonstrate that the proposed auction mechanism can represent almost any strategy-proof auction mechanism, and outperforms the auction mechanisms wildly used in the correlated value settings.

cs.GT↗

arXiv2022

BSA -- Bi-Stiffness Actuation for optimally exploiting intrinsic compliance and inertial coupling effects in elastic joint robots

Compliance in actuation has been exploited to generate highly dynamic maneuvers such as throwing that take advantage of the potential energy stored in joint springs. However, the energy storage and release could not be well-timed yet. On the contrary, for multi-link systems, the natural system dynamics might even work against the actual goal. With the introduction of variable stiffness actuators, this problem has been partially addressed. With a suitable optimal control strategy, the approximate decoupling of the motor from the link can be achieved to maximize the energy transfer into the distal link prior to launch. However, such continuous stiffness variation is complex and typically leads to oscillatory swing-up motions instead of clear launch sequences. To circumvent this issue, we investigate decoupling for speed maximization with a dedicated novel actuator concept denoted Bi-Stiffness Actuation. With this, it is possible to fully decouple the link from the joint mechanism by a switch-and-hold clutch and simultaneously keep the elastic energy stored. We show that with this novel paradigm, it is not only possible to reach the same optimal performance as with power-equivalent variable stiffness actuation, but even directly control the energy transfer timing. This is a major step forward compared to previous optimal control approaches, which rely on optimizing the full time-series control input.

cs.RO↗

arXiv2022

Critical quantum sensing based on the Jaynes-Cummings model with a squeezing drive

Quantum sensing improves the accuracy of measurements of relevant parameters by exploiting the unique properties of quantum systems. The divergent susceptibility of physical systems near a critical point for quantum phase transition enables criticality-enhanced quantum sensing. The quantum Rabi model (QRM), composed of a single qubit coupled to a single bosonic field, represents a good candidate for realizing such critical enhancement for its simplicity, but it is experimentally challenging to achieve the ultrastrong qubit-field coupling required to realize the critical phenomena. In this work, we explore an alternative to construct the analog of the QRM for the sensing, exploiting the criticality appearing in the Jaynes-Cummings (JC) model whose bosonic field is parametrically driven, not necessitating the ultrastrong coupling condition thus to some extent relaxing the requirement for the practical implementation.

quant-ph↗

arXiv2022

evoML Yellow Paper: Evolutionary AI and Optimisation Studio

Machine learning model development and optimisation can be a rather cumbersome and resource-intensive process. Custom models are often more difficult to build and deploy, and they require infrastructure and expertise which are often costly to acquire and maintain. Machine learning product development lifecycle must take into account the need to navigate the difficulties of developing and deploying machine learning models. evoML is an AI-powered tool that provides automated functionalities in machine learning model development, optimisation, and model code optimisation. Core functionalities of evoML include data cleaning, exploratory analysis, feature analysis and generation, model optimisation, model evaluation, model code optimisation, and model deployment. Additionally, a key feature of evoML is that it embeds code and model optimisation into the model development process, and includes multi-objective optimisation capabilities.

cs.AI↗

arXiv2022

Ultra-Low-Frequency Radio Astronomy Observations from a Selenocentric Orbit: first results of the Longjiang-2 experiment

This paper introduces the first results of observations with the Ultra-Long-Wavelength (ULW) -- Low Frequency Interferometer and Spectrometer (LFIS) on board the selenocentric satellite Longjiang-2. We present a brief description of the satellite and focus on the LFIS payload. The in-orbit commissioning confirmed a reliable operational status of the instrumentation. We also present results of a transition observation, which offers unique measurements on several novel aspects. We estimate the RFI suppression required for such a radio astronomy instrumentation at the Moon distances from Earth to be of the order of 80 dB. We analyse a method of separating Earth- and satellite-originated radio frequency interference (RFI). It is found that the RFI level at frequencies lower than a few MHz is smaller than the receiver noise floor.

astro-ph.IM↗

arXiv2022

Detection of brain activations induced by naturalistic stimuli in a pseudo model-driven way

Naturalistic fMRI has been suggested to be a powerful alternative for investigations of human brain function. Stimulus-induced activation has been playing an essential role in fMRI-based brain function analyses. Due to the complexity of the stimuli, however, detection of activations induced by naturalistic stimuli (AINSs) has been a tricky problem, as AINS cannot be detected simply in a model-driven way. In this study, we proposed a method to detect AINS in a pseudo model-driven way. Inspired by the strategy of utilizing the commonalities among the brains exposed to the same stimuli for inter-subject correlation analysis, we established response models for one subject by averaging the fMRI signals across several other subjects, and then detected AINSs of the subject using general linear model. We evaluated the effectiveness of AINS with both statistical and predictive analyses on individual differences in sex and intelligence quotient (IQ), based on the four movie fMRI runs included in the Human Connectome Project dataset. The results indicate that AINS is not only sensitive to sex- and IQ-related differences, but also specific enough to decode individuals' sex and IQ. Specifically, activations in brain regions associated with visual-spatial processing were observed to be consistently stronger in the males, and individuals with higher IQ exhibited consistently stronger activations in regions within the visual and the default mode networks. Predictions of individuals' sex and IQ were significantly better than those based on random labels (P < 0.005). Taken together, AINS advanced in this study can be an effective evaluation of human brain function. The conceptual simplicity and easy application of its detection may make AINS a favorable choice for future brain function analyses and personalized medicine based on naturalistic fMRI.

q-bio.NC↗

arXiv2022

Utility Maximizer or Value Maximizer: Mechanism Design for Mixed Bidders in Online Advertising

Digital advertising constitutes one of the main revenue sources for online platforms. In recent years, some advertisers tend to adopt auto-bidding tools to facilitate advertising performance optimization, making the classical \emph{utility maximizer} model in auction theory not fit well. Some recent studies proposed a new model, called \emph{value maximizer}, for auto-bidding advertisers with return-on-investment (ROI) constraints. However, the model of either utility maximizer or value maximizer could only characterize partial advertisers in real-world advertising platforms. In a mixed environment where utility maximizers and value maximizers coexist, the truthful ad auction design would be challenging since bidders could manipulate both their values and affiliated classes, leading to a multi-parameter mechanism design problem. In this work, we address this issue by proposing a payment rule which combines the corresponding ones in classical VCG and GSP mechanisms in a novel way. Based on this payment rule, we propose a truthful auction mechanism with an approximation ratio of $2$ on social welfare, which is close to the lower bound of at least $\frac{5}{4}$ that we also prove. The designed auction mechanism is a generalization of VCG for utility maximizers and GSP for value maximizers.

cs.GT↗

arXiv2022

One-Time Model Adaptation to Heterogeneous Clients: An Intra-Client and Inter-Image Attention Design

The mainstream workflow of image recognition applications is first training one global model on the cloud for a wide range of classes and then serving numerous clients, each with heterogeneous images from a small subset of classes to be recognized. From the cloud-client discrepancies on the range of image classes, the recognition model is desired to have strong adaptiveness, intuitively by concentrating the focus on each individual client's local dynamic class subset, while incurring negligible overhead. In this work, we propose to plug a new intra-client and inter-image attention (ICIIA) module into existing backbone recognition models, requiring only one-time cloud-based training to be client-adaptive. In particular, given a target image from a certain client, ICIIA introduces multi-head self-attention to retrieve relevant images from the client's historical unlabeled images, thereby calibrating the focus and the recognition result. Further considering that ICIIA's overhead is dominated by linear projection, we propose partitioned linear projection with feature shuffling for replacement and allow increasing the number of partitions to dramatically improve efficiency without scarifying too much accuracy. We finally evaluate ICIIA using 3 different recognition tasks with 9 backbone models over 5 representative datasets. Extensive evaluation results demonstrate the effectiveness and efficiency of ICIIA. Specifically, for ImageNet-1K with the backbone models of MobileNetV3-L and Swin-B, ICIIA can improve the testing accuracy to 83.37% (+8.11%) and 88.86% (+5.28%), while adding only 1.62% and 0.02% of FLOPs, respectively.

cs.CV↗

arXiv2022

On-Device Model Fine-Tuning with Label Correction in Recommender Systems

To meet the practical requirements of low latency, low cost, and good privacy in online intelligent services, more and more deep learning models are offloaded from the cloud to mobile devices. To further deal with cross-device data heterogeneity, the offloaded models normally need to be fine-tuned with each individual user's local samples before being put into real-time inference. In this work, we focus on the fundamental click-through rate (CTR) prediction task in recommender systems and study how to effectively and efficiently perform on-device fine-tuning. We first identify the bottleneck issue that each individual user's local CTR (i.e., the ratio of positive samples in the local dataset for fine-tuning) tends to deviate from the global CTR (i.e., the ratio of positive samples in all the users' mixed datasets on the cloud for training out the initial model). We further demonstrate that such a CTR drift problem makes on-device fine-tuning even harmful to item ranking. We thus propose a novel label correction method, which requires each user only to change the labels of the local samples ahead of on-device fine-tuning and can well align the locally prior CTR with the global CTR. The offline evaluation results over three datasets and five CTR prediction models as well as the online A/B testing results in Mobile Taobao demonstrate the necessity of label correction in on-device fine-tuning and also reveal the improvement over cloud-based learning without fine-tuning.

cs.IR↗

arXiv2022

Mining Causality from Continuous-time Dynamics Models: An Application to Tsunami Forecasting

Continuous-time dynamics models, such as neural ordinary differential equations, have enabled the modeling of underlying dynamics in time-series data and accurate forecasting. However, parameterization of dynamics using a neural network makes it difficult for humans to identify causal structures in the data. In consequence, this opaqueness hinders the use of these models in the domains where capturing causal relationships carries the same importance as accurate predictions, e.g., tsunami forecasting. In this paper, we address this challenge by proposing a mechanism for mining causal structures from continuous-time models. We train models to capture the causal structure by enforcing sparsity in the weights of the input layers of the dynamics models. We first verify the effectiveness of our method in the scenario where the exact causal-structures of time-series are known as a priori. We next apply our method to a real-world problem, namely tsunami forecasting, where the exact causal-structures are difficult to characterize. Experimental results show that the proposed method is effective in learning physically-consistent causal relationships while achieving high forecasting accuracy.

cs.LG↗

arXiv2022

SecretGen: Privacy Recovery on Pre-Trained Models via Distribution Discrimination

Transfer learning through the use of pre-trained models has become a growing trend for the machine learning community. Consequently, numerous pre-trained models are released online to facilitate further research. However, it raises extensive concerns on whether these pre-trained models would leak privacy-sensitive information of their training data. Thus, in this work, we aim to answer the following questions: "Can we effectively recover private information from these pre-trained models? What are the sufficient conditions to retrieve such sensitive information?" We first explore different statistical information which can discriminate the private training distribution from other distributions. Based on our observations, we propose a novel private data reconstruction framework, SecretGen, to effectively recover private information. Compared with previous methods which can recover private data with the ground true prediction of the targeted recovery instance, SecretGen does not require such prior knowledge, making it more practical. We conduct extensive experiments on different datasets under diverse scenarios to compare SecretGen with other baselines and provide a systematic benchmark to better understand the impact of different auxiliary information and optimization operations. We show that without prior knowledge about true class prediction, SecretGen is able to recover private data with similar performance compared with the ones that leverage such prior knowledge. If the prior knowledge is given, SecretGen will significantly outperform baseline methods. We also propose several quantitative metrics to further quantify the privacy vulnerability of pre-trained models, which will help the model selection for privacy-sensitive applications. Our code is available at: https://github.com/AI-secure/SecretGen.

cs.LG↗

arXiv2022

Deep learning-based identification of sub-nuclear structures in FIB-SEM images

Three-dimensional volumetric imaging of cells allows for in situ visualization, thus preserving contextual insights into cellular processes. Despite recent advances in machine learning methods, morphological analysis of sub-nuclear structures have proven challenging due to both the shallow contrast profile and the technical limitation in feature detection. Here, we present a convolutional neural network, supervised deep learning-based approach which can identify sub-nuclear structures with 90% accuracy. We develop and apply this model to C. elegans gonads imaged using focused ion beam milling combined with scanning electron microscopy resulting in the accurate identification and segmentation of all sub-nuclear structures including entire chromosomes. We discuss in depth the architecture, parameterization, and optimization of the deep learning model, as well as provide evaluation metrics to assess the quality of the network prediction. Lastly, we highlight specific aspects of the model that can be optimized for its broad application to other volumetric imaging data as well as in situ cryo-electron tomography.

q-bio.QM↗

arXiv2022

Wafer-scale epitaxial growth of the thickness-controllable van der Waals ferromagnet CrTe2 for reliable magnetic memory applications

To harness the intriguing properties of two-dimensional van der Waals (vdW) ferromagnets (FMs) for versatile applications, the key challenge lies in the reliable material synthesis for scalable device production. Here, we demonstrate the epitaxial growth of single-crystalline 1T-CrTe2 thin films on 2-inch sapphire substrates. Benefiting from the uniform surface energy of the dangling bond-free Al2O3(0001) surface, the layer-by-layer vdW growth mode is observed right from the initial growth stage, which warrants precise control of the sample thickness and atomically smooth surface morphology across the entire wafer. Moreover, the presence of the Coulomb interaction at the CrTe2/Al2O3 interface serves as an effective tuning parameter to tailor the anomalous Hall response, and the structural optimization of the CrTe2-based spin-orbit torque device leads to a substantial switching power reduction by 54%. Our results may lay out a general framework for the design of energy-efficient spintronics based on configurable vdW FMs.

cond-mat.mtrl-sci↗

arXiv2022

Angular topological superfluid and topological vortex in an ultracold Fermi gas

We show that pairing in an ultracold Fermi gas under spin-orbital-angular-momentum coupling (SOAMC) can acquire topological characters encoded in the quantized angular degrees of freedom. The resulting topological superfluid is the angular analog of its counterpart in a one-dimensional Fermi gas with spin-orbit coupling, but characterized by a Zak phase defined in the angular-momentum space. Upon tuning the SOAMC parameters, a topological phase transition occurs, which is accompanied by the closing of the quasiparticle excitation gap. Remarkably, a topological vortex state can also be stabilized by deforming the Fermi surface, which is topologically non-trivial in both the coordinate and angular-momentum space, offering interesting potentials for applications in quantum information and quantum control. We discuss how the topological phase transition and the exotic vortex state can be detected experimentally.

cond-mat.quant-gas↗