arXiv subjects

Haotian Jiang

Publications and source records attributed to Haotian Jiang.

At least 37 records · Page 2Linked to original sources

arXiv2025

MITracker: Multi-View Integration for Visual Object Tracking

Multi-view object tracking (MVOT) offers promising solutions to challenges such as occlusion and target loss, which are common in traditional single-view tracking. However, progress has been limited by the lack of comprehensive multi-view datasets and effective cross-view integration methods. To overcome these limitations, we compiled a Multi-View object Tracking (MVTrack) dataset of 234K high-quality annotated frames featuring 27 distinct objects across various scenes. In conjunction with this dataset, we introduce a novel MVOT method, Multi-View Integration Tracker (MITracker), to efficiently integrate multi-view object features and provide stable tracking outcomes. MITracker can track any object in video frames of arbitrary length from arbitrary viewpoints. The key advancements of our method over traditional single-view approaches come from two aspects: (1) MITracker transforms 2D image features into a 3D feature volume and compresses it into a bird's eye view (BEV) plane, facilitating inter-view information fusion; (2) we propose an attention mechanism that leverages geometric information from fused 3D feature volume to refine the tracking results at each view. MITracker outperforms existing methods on the MVTrack and GMTD datasets, achieving state-of-the-art performance. The code and the new dataset will be available at https://mii-laboratory.github.io/MITracker/.

cs.CV↗

arXiv2025

DCIM-AVSR : Efficient Audio-Visual Speech Recognition via Dual Conformer Interaction Module

Speech recognition is the technology that enables machines to interpret and process human speech, converting spoken language into text or commands. This technology is essential for applications such as virtual assistants, transcription services, and communication tools. The Audio-Visual Speech Recognition (AVSR) model enhances traditional speech recognition, particularly in noisy environments, by incorporating visual modalities like lip movements and facial expressions. While traditional AVSR models trained on large-scale datasets with numerous parameters can achieve remarkable accuracy, often surpassing human performance, they also come with high training costs and deployment challenges. To address these issues, we introduce an efficient AVSR model that reduces the number of parameters through the integration of a Dual Conformer Interaction Module (DCIM). In addition, we propose a pre-training method that further optimizes model performance by selectively updating parameters, leading to significant improvements in efficiency. Unlike conventional models that require the system to independently learn the hierarchical relationship between audio and visual modalities, our approach incorporates this distinction directly into the model architecture. This design enhances both efficiency and performance, resulting in a more practical and effective solution for AVSR tasks.

eess.AS↗

arXiv2025

Approximation Rate of the Transformer Architecture for Sequence Modeling

The Transformer architecture is widely applied in sequence modeling applications, yet the theoretical understanding of its working principles remains limited. In this work, we investigate the approximation rate for single-layer Transformers with one head. We consider a class of non-linear relationships and identify a novel notion of complexity measures to establish an explicit Jackson-type approximation rate estimate for the Transformer. This rate reveals the structural properties of the Transformer and suggests the types of sequential relationships it is best suited for approximating. In particular, the results on approximation rates enable us to concretely analyze the differences between the Transformer and classical sequence modeling methods, such as recurrent neural networks.

cs.LG↗

arXiv2024

A Geometric Perspective on the Injective Norm of Sums of Random Tensors

Matrix concentration inequalities, intimately connected to the Non-Commutative Khintchine inequality, have been an important tool in both applied and pure mathematics. We study tensor versions of these inequalities, and establish non-asymptotic inequalities for the $\ell_p$ injective norm of random tensors with correlated entries. In certain regimes of $p$ and the tensor order, our tensor concentration inequalities are nearly optimal in their dimension dependencies. We illustrate our result with applications to problems including structured models of random tensors and matrices, tensor PCA, and connections to lower bounds in coding theory. Our techniques are based on covering number estimates as opposed to operator theoretic tools, which also provide a geometric proof of a weaker version of the Non-Commutative Khintchine inequality, motivated by a question of Talagrand.

math.PR↗

arXiv2024

Quasi-Monte Carlo Beyond Hardy-Krause

The classical approaches to numerically integrating a function $f$ are Monte Carlo (MC) and quasi-Monte Carlo (QMC) methods. MC methods use random samples to evaluate $f$ and have error $O(σ(f)/\sqrt{n})$, where $σ(f)$ is the standard deviation of $f$. QMC methods are based on evaluating $f$ at explicit point sets with low discrepancy, and as given by the classical Koksma-Hlawka inequality, they have error $\widetilde{O}(σ_{\mathsf{HK}}(f)/n)$, where $σ_{\mathsf{HK}}(f)$ is the variation of $f$ in the sense of Hardy and Krause. These two methods have distinctive advantages and shortcomings, and a fundamental question is to find a method that combines the advantages of both. In this work, we give a simple randomized algorithm that produces QMC point sets with the following desirable features: (1) It achieves substantially better error than given by the classical Koksma-Hlawka inequality. In particular, it has error $\widetilde{O}(σ_{\mathsf{SO}}(f)/n)$, where $σ_{\mathsf{SO}}(f)$ is a new measure of variation that we introduce, which is substantially smaller than the Hardy-Krause variation. (2) The algorithm only requires random samples from the underlying distribution, which makes it as flexible as MC. (3) It automatically achieves the best of both MC and QMC (and the above improvement over Hardy-Krause variation) in an optimal way. (4) The algorithm is extremely efficient, with an amortized $\widetilde{O}(1)$ runtime per sample. Our method is based on the classical transference principle in geometric discrepancy, combined with recent algorithmic innovations in combinatorial discrepancy that besides producing low-discrepancy colorings, also guarantee certain subgaussian properties. This allows us to bypass several limitations of previous works in bridging the gap between MC and QMC methods and go beyond the Hardy-Krause variation.

cs.DS↗

arXiv2024

Differentially Private Synthetic Data via Foundation Model APIs 2: Text

Text data has become extremely valuable due to the emergence of machine learning algorithms that learn from it. A lot of high-quality text data generated in the real world is private and therefore cannot be shared or used freely due to privacy concerns. Generating synthetic replicas of private text data with a formal privacy guarantee, i.e., differential privacy (DP), offers a promising and scalable solution. However, existing methods necessitate DP finetuning of large language models (LLMs) on private data to generate DP synthetic data. This approach is not viable for proprietary LLMs (e.g., GPT-3.5) and also demands considerable computational resources for open-source LLMs. Lin et al. (2024) recently introduced the Private Evolution (PE) algorithm to generate DP synthetic images with only API access to diffusion models. In this work, we propose an augmented PE algorithm, named Aug-PE, that applies to the complex setting of text. We use API access to an LLM and generate DP synthetic text without any model training. We conduct comprehensive experiments on three benchmark datasets. Our results demonstrate that Aug-PE produces DP synthetic text that yields competitive utility with the SOTA DP finetuning baselines. This underscores the feasibility of relying solely on API access of LLMs to produce high-quality DP synthetic texts, thereby facilitating more accessible routes to privacy-preserving LLM applications. Our code and data are available at https://github.com/AI-secure/aug-pe.

cs.CL↗

arXiv2024

Sparse Submodular Function Minimization

In this paper we study the problem of minimizing a submodular function $f : 2^V \rightarrow \mathbb{R}$ that is guaranteed to have a $k$-sparse minimizer. We give a deterministic algorithm that computes an additive $ε$-approximate minimizer of such $f$ in $\widetilde{O}(\mathsf{poly}(k) \log(|f|/ε))$ parallel depth using a polynomial number of queries to an evaluation oracle of $f$, where $|f| = \max_{S \subseteq V} |f(S)|$. Further, we give a randomized algorithm that computes an exact minimizer of $f$ with high probability using $\widetilde{O}(|V| \cdot \mathsf{poly}(k))$ queries and polynomial time. When $k = \widetilde{O}(1)$, our algorithms use either nearly-constant parallel depth or a nearly-linear number of evaluation oracle queries. All previous algorithms for this problem either use $Ω(|V|)$ parallel depth or $Ω(|V|^2)$ queries. In contrast to state-of-the-art weakly-polynomial and strongly-polynomial time algorithms for SFM, our algorithms use first-order optimization methods, e.g., mirror descent and follow the regularized leader. We introduce what we call {\em sparse dual certificates}, which encode information on the structure of sparse minimizers, and both our parallel and sequential algorithms provide new algorithmic tools for allowing first-order optimization methods to efficiently compute them. Correspondingly, our algorithm does not invoke fast matrix multiplication or general linear system solvers and in this sense is more combinatorial than previous state-of-the-art methods.

cs.DS↗

arXiv2023

Convex Minimization with Integer Minima in $\widetilde O(n^4)$ Time

Given a convex function $f$ on $\mathbb{R}^n$ with an integer minimizer, we show how to find an exact minimizer of $f$ using $O(n^2 \log n)$ calls to a separation oracle and $O(n^4 \log n)$ time. The previous best polynomial time algorithm for this problem given in [Jiang, SODA 2021, JACM 2022] achieves $O(n^2\log\log n/\log n)$ oracle complexity. However, the overall runtime of Jiang's algorithm is at least $\widetildeΩ(n^8)$, due to expensive sub-routines such as the Lenstra-Lenstra-Lovász (LLL) algorithm [Lenstra, Lenstra, Lovász, Math. Ann. 1982] and random walk based cutting plane method [Bertsimas, Vempala, JACM 2004]. Our significant speedup is obtained by a nontrivial combination of a faster version of the LLL algorithm due to [Neumaier, Stehlé, ISSAC 2016] that gives similar guarantees, the volumetric center cutting plane method (CPM) by [Vaidya, FOCS 1989] and its fast implementation given in [Jiang, Lee, Song, Wong, STOC 2020]. For the special case of submodular function minimization (SFM), our result implies a strongly polynomial time algorithm for this problem using $O(n^3 \log n)$ calls to an evaluation oracle and $O(n^4 \log n)$ additional arithmetic operations. Both the oracle complexity and the number of arithmetic operations of our more general algorithm are better than the previous best-known runtime algorithms for this specific problem given in [Lee, Sidford, Wong, FOCS 2015] and [Dadush, Végh, Zambelli, SODA 2018, MOR 2021].

cs.DS↗

arXiv2023

Parallel Submodular Function Minimization

We consider the parallel complexity of submodular function minimization (SFM). We provide a pair of methods which obtain two new query versus depth trade-offs a submodular function defined on subsets of $n$ elements that has integer values between $-M$ and $M$. The first method has depth $2$ and query complexity $n^{O(M)}$ and the second method has depth $\widetilde{O}(n^{1/3} M^{2/3})$ and query complexity $O(\mathrm{poly}(n, M))$. Despite a line of work on improved parallel lower bounds for SFM, prior to our work the only known algorithms for parallel SFM either followed from more general methods for sequential SFM or highly-parallel minimization of convex $\ell_2$-Lipschitz functions. Interestingly, to obtain our second result we provide the first highly-parallel algorithm for minimizing $\ell_\infty$-Lipschitz function over the hypercube which obtains near-optimal depth for obtaining constant accuracy.

cs.DS↗

arXiv2023

Forward and Inverse Approximation Theory for Linear Temporal Convolutional Networks

We present a theoretical analysis of the approximation properties of convolutional architectures when applied to the modeling of temporal sequences. Specifically, we prove an approximation rate estimate (Jackson-type result) and an inverse approximation theorem (Bernstein-type result), which together provide a comprehensive characterization of the types of sequential relationships that can be efficiently captured by a temporal convolutional architecture. The rate estimate improves upon a previous result via the introduction of a refined complexity measure, whereas the inverse approximation theorem is new.

cs.LG↗

arXiv2023

Learning across Data Owners with Joint Differential Privacy

In this paper, we study the setting in which data owners train machine learning models collaboratively under a privacy notion called joint differential privacy [Kearns et al., 2018]. In this setting, the model trained for each data owner $j$ uses $j$'s data without privacy consideration and other owners' data with differential privacy guarantees. This setting was initiated in [Jain et al., 2021] with a focus on linear regressions. In this paper, we study this setting for stochastic convex optimization (SCO). We present an algorithm that is a variant of DP-SGD [Song et al., 2013; Abadi et al., 2016] and provides theoretical bounds on its population loss. We compare our algorithm to several baselines and discuss for what parameter setups our algorithm is more preferred. We also empirically study joint differential privacy in the multi-class classification problem over two public datasets. Our empirical findings are well-connected to the insights from our theoretical results.

cs.LG↗

arXiv2023

A Brief Survey on the Approximation Theory for Sequence Modelling

We survey current developments in the approximation theory of sequence modelling in machine learning. Particular emphasis is placed on classifying existing results for various model architectures through the lens of classical approximation paradigms, and the insights one can gain from these results. We also outline some future research directions towards building a theory of sequence modelling.

cs.LG↗

arXiv2022

Saiyan: Design and Implementation of a Low-power Demodulator for LoRa Backscatter Systems

The radio range of backscatter systems continues growing as new wireless communication primitives are continuously invented. Nevertheless, both the bit error rate and the packet loss rate of backscatter signals increase rapidly with the radio range, thereby necessitating the cooperation between the access point and the backscatter tags through a feedback loop. Unfortunately, the low-power nature of backscatter tags limits their ability to demodulate feedback signals from a remote access point and scales down to such circumstances. This paper presents Saiyan, an ultra-low-power demodulator for long-range LoRa backscatter systems. With Saiyan, a backscatter tag can demodulate feedback signals from a remote access point with moderate power consumption and then perform an immediate packet retransmission in the presence of packet loss. Moreover, Saiyan enables rate adaption and channel hopping-two PHY-layer operations that are important to channel efficiency yet unavailable on long-range backscatter systems. We prototype Saiyan on a two-layer PCB board and evaluate its performance in different environments. Results show that Saiyan achieves 5 gain on the demodulation range, compared with state-of-the-art systems. Our ASIC simulation shows that the power consumption of Saiyan is around 93.2 uW. Code and hardware schematics can be found at: https://github.com/ZangJac/Saiyan.

cs.NI↗

arXiv2022

Efficient Ambient LoRa Backscatter with On-Off Keying Modulation

Backscatter communication holds potential for ubiquitous and low-cost connectivity among low-power IoT devices. To avoid interference between the carrier signal and the backscatter signal, recent works propose a frequency-shifting technique to separate these two signals in the frequency domain. Such proposals, however, have to occupy the precious wireless spectrum that is already overcrowded, and increase the power, cost, and complexity of the backscatter tag. In this paper, we revisit the classic ON-OFF Keying (OOK) modulation and propose Aloba, a backscatter system that takes the ambient LoRa transmissions as the excitation and piggybacks the in-band OOK modulated signals over the LoRa transmissions. Our design enables the backsactter signal to work in the same frequency band of the carrier signal, meanwhile achieving flexible data rate at different transmission range. The key contributions of Aloba include: (1) the design of a low-power backscatter tag that can pick up the ambient LoRa signals from other signals. (2) a novel decoding algorithm to demodulate both the carrier signal and the backscatter signal from their superposition. We further adopt link coding mechanism and interleave operation to enhance the reliability of backscatter signal decoding. We implement Aloba and conduct head-to-head comparison with the state-of-the-art LoRa backscatter system PLoRa in various settings. The experiment results show Aloba can achieve 199.4 Kbps data rate at various distances, 52.4 times higher than PLoRa.

cs.NI↗

arXiv2022

Minimizing Convex Functions with Rational Minimizers

Given a separation oracle $\mathsf{SO}$ for a convex function $f$ defined on $\mathbb{R}^n$ that has an integral minimizer inside a box with radius $R$, we show how to find an exact minimizer of $f$ using at most (a) $O(n (n \log \log (n)/\log (n) + \log(R)))$ calls to $\mathsf{SO}$ and $\mathsf{poly}(n, \log(R))$ arithmetic operations, or (b) $O(n \log(nR))$ calls to $\mathsf{SO}$ and $\exp(O(n)) \cdot \mathsf{poly}(\log(R))$ arithmetic operations. When the set of minimizers of $f$ has integral extreme points, our algorithm outputs an integral minimizer of $f$. This improves upon the previously best oracle complexity of $O(n^2 (n + \log(R)))$ for polynomial time algorithms and $O(n^2\log(nR))$ for exponential time algorithms obtained by [Grötschel, Lovász and Schrijver, Prog. Comb. Opt. 1984, Springer 1988] over thirty years ago. Our improvement on Grötschel, Lovász and Schrijver's result generalizes to the setting where the set of minimizers of $f$ is a rational polyhedron with bounded vertex complexity. For the Submodular Function Minimization problem, our result immediately implies a strongly polynomial algorithm that makes at most $O(n^3 \log \log (n)/\log (n))$ calls to an evaluation oracle, and an exponential time algorithm that makes at most $O(n^2 \log(n))$ calls to an evaluation oracle. These improve upon the previously best $O(n^3 \log^2(n))$ oracle complexity for strongly polynomial algorithms given in [Lee, Sidford and Wong, FOCS 2015] and [Dadush, Végh and Zambelli, SODA 2018], and an exponential time algorithm with oracle complexity $O(n^3 \log(n))$ given in the former work. Our result is achieved via a reduction to the Shortest Vector Problem in lattices. We analyze its oracle complexity using a potential function that simultaneously captures the size of the search set and the density of the lattice.

cs.DS↗

arXiv2022

Resolving Matrix Spencer Conjecture Up to Poly-logarithmic Rank

We give a simple proof of the matrix Spencer conjecture up to poly-logarithmic rank: given symmetric $d \times d$ matrices $A_1,\ldots,A_n$ each with $\|A_i\|_{\mathsf{op}} \leq 1$ and rank at most $n/\log^3 n$, one can efficiently find $\pm 1$ signs $x_1,\ldots,x_n$ such that their signed sum has spectral norm $\|\sum_{i=1}^n x_i A_i\|_{\mathsf{op}} = O(\sqrt{n})$. This result also implies a $\log n - Ω( \log \log n)$ qubit lower bound for quantum random access codes encoding $n$ classical bits with advantage $\gg 1/\sqrt{n}$. Our proof uses the recent refinement of the non-commutative Khintchine inequality in [Bandeira, Boedihardjo, van Handel, 2022] for random matrices with correlated Gaussian entries.

cs.DS↗

arXiv2022

Decomposable Non-Smooth Convex Optimization with Nearly-Linear Gradient Oracle Complexity

Many fundamental problems in machine learning can be formulated by the convex program \[ \min_{θ\in R^d}\ \sum_{i=1}^{n}f_{i}(θ), \] where each $f_i$ is a convex, Lipschitz function supported on a subset of $d_i$ coordinates of $θ$. One common approach to this problem, exemplified by stochastic gradient descent, involves sampling one $f_i$ term at every iteration to make progress. This approach crucially relies on a notion of uniformity across the $f_i$'s, formally captured by their condition number. In this work, we give an algorithm that minimizes the above convex formulation to $ε$-accuracy in $\widetilde{O}(\sum_{i=1}^n d_i \log (1 /ε))$ gradient computations, with no assumptions on the condition number. The previous best algorithm independent of the condition number is the standard cutting plane method, which requires $O(nd \log (1/ε))$ gradient computations. As a corollary, we improve upon the evaluation oracle complexity for decomposable submodular minimization by Axiotis et al. (ICML 2021). Our main technical contribution is an adaptive procedure to select an $f_i$ term at every iteration via a novel combination of cutting-plane and interior-point methods.

math.OC↗

arXiv2022

Improved Lower Bounds for Submodular Function Minimization

We provide a generic technique for constructing families of submodular functions to obtain lower bounds for submodular function minimization (SFM). Applying this technique, we prove that any deterministic SFM algorithm on a ground set of $n$ elements requires at least $Ω(n \log n)$ queries to an evaluation oracle. This is the first super-linear query complexity lower bound for SFM and improves upon the previous best lower bound of $2n$ given by [Graur et al., ITCS 2020]. Using our construction, we also prove that any (possibly randomized) parallel SFM algorithm, which can make up to $\mathsf{poly}(n)$ queries per round, requires at least $Ω(n / \log n)$ rounds to minimize a submodular function. This improves upon the previous best lower bound of $\tildeΩ(n^{1/3})$ rounds due to [Chakrabarty et al., FOCS 2021], and settles the parallel complexity of query-efficient SFM up to logarithmic factors due to a recent advance in [Jiang, SODA 2021].

cs.DS↗