Searcharxiv⌕ Search

arXiv subjects

Hanzhi Wang

Publications and source records attributed to Hanzhi Wang.

At least 19 recordsLinked to original sources

Estimating Random-Walk Probabilities in Directed Graphs

We study discounted random walks in directed graphs. In each step, the walk either terminates with a constant probability $α$, or proceeds to a random out-neighbor. Our goal is to estimate the probability $π(s, t)$ that a discounted random walk starting from $s$ terminates at $t$. This probability is also known as the Personalized PageRank (PPR) score, which measures the relevance of $t$ to $s$, for instance, when $s$ and $t$ are web pages on the Internet. We aim to estimate $π(s, t)$ within a constant relative error with constant probability. A variety of algorithms have been developed for several problem variants, such as single-pair, single-source, single-target, and single-node estimation, under both worst-case and average-case settings, and for different combinations of allowed graph queries. However, in many important cases, there remain polynomial gaps between known upper and lower bounds. In this paper, we establish tight bounds for all problem variants and query combinations, closing all existing gaps in both the worst-case and average-case settings. We provide tight (up to logarithmic factors) lower bounds, showing that for all but one query combination, existing algorithms are already optimal. For the remaining case, we design a novel algorithm that matches our new lower bound up to logarithmic factors, thereby nearly achieving optimality. This is the first algorithm to exploit this specific query combination. It uses a new randomized bidirectional framework that combines randomized backward propagation with selective Monte Carlo estimation.

cs.DS↗

A Binary Search Tree with the Unified Bound

We give a deterministic online binary search tree (BST) that satisfies the unified bound of Iacono [SODA 2001], which subsumes the working-set and dynamic-finger bounds: it serves every sequence $X$ of $m$ searches on $n$ keys at total cost $O(m+\mathrm{UB}(X))$, and every single search at cost $O(\log n)$. Previously, no BST was known to satisfy the unified bound, even offline. As a consequence, the dynamic optimality conjecture implies Iacono's conjecture that splay trees satisfy the unified bound. We also show that the unified bound is $O((m+n)(1+\log c_σ))$ on every sequence of searches that avoids a fixed permutation $σ$, where $c_σ$ is the Füredi-Hajnal limit of $σ$. Hence every BST that satisfies the unified bound, including ours, serves such sequences at this cost. For permutations this bound is tight, and it improves the previous upper bound $O(nc_σ^2)$ on the optimal cost, due to Berendsohn, Kozma and Opler [STOC 2024]. Finally, we extend the tree to support insertions and deletions: a sequence $X$ of $m$ operations, starting from $n_0$ keys, costs $O(n_0+m+\mathrm{UB}(X))$ in total, where $\mathrm{UB}$ is now the extension of the unified bound to insertions and deletions by Bădoiu, Cole, Demaine and Iacono [TCS 2007]. Hence, if splay trees are dynamically optimal for searches, insertions and deletions, then they serve every such sequence at this total cost as well.

cs.DS↗

Instance-Optimality of Single-Pair Personalized PageRank

We consider discounted random walks in a directed simple graph $G=(V,E)$. Such a walk terminates at each step with a constant probability $α\in (0,1)$ and otherwise continues to a uniform random out-neighbor. Given a pair $(s,t)\in V^2$, we denote by $π(s,t)$ the probability of such a discounted random walk starting at $s$ terminates at $t$. This probability is referred to as the Personalized PageRank (PPR) of $t$ with respect to $s$. The problem of estimating $π(s,t)$ within a constant multiplicative factor with constant success probability for a given vertex pair $(s,t)$ is called the single-pair PPR estimation problem. We investigate the instance-optimality of single-pair PPR estimation in directed graphs. We present a new algorithm that aims at instance-optimal running time up to polylogarithmic factors. In other words, no correct algorithm can run faster than our algorithm on any valid graph by more than a polylogarithmic factor. We show that our algorithm is instance-optimal in the standard adjacency-list query model if $n$ is not known (as is often the case for most of the large networks to which PPR is applied). When $n$ is known, our algorithm is still instance-optimal for sparse graphs. For denser graphs, when $n$ is known, we show that no algorithm can achieve general instance-optimality. For the multiplicative approximation of PPR, we also improve the worst-case complexity from $O(mn\log n)$ to the near-optimal $O(m\log n)$. This is not only for the single-pair discussed above, but also for the single-source and single-target variants of the PPR estimation problem.

cs.DS↗

Faster Algorithms for Disjoint Paths in DAGs and Two Disjoint Shortest Paths

For $k$ vertex-disjoint paths with prescribed endpoints in a directed acyclic graph (DAG), the $O(mn^{k-1})$-time pebbling algorithm of Fortune, Hopcroft, and Wyllie [TCS 1980] has remained the fastest for every $k\geq3$ for over four decades, even when the paths must be shortest. We break this bound for every $k\geq3$ with randomized algorithms running in expected $O_k(m+n^{k-3+ω+\varepsilon})$ time for every $\varepsilon>0$, where $ω<2.372$ is the matrix multiplication exponent; this is $O_k(m+n^{k-0.62})$. For two disjoint shortest paths, we find the paths as fast as we detect them: in expected linear time on DAGs and on undirected graphs with polynomially bounded integer weights, and in expected $O(\min\{mn,n^{ω+\varepsilon}\})$ time on directed graphs whose cycles have positive weight, which on dense graphs is $o(mn)$ even for detection. All algorithms evaluate a polynomial whose monomials are the solutions: a new first-collision circuit speeds up the evaluation in DAGs, and a correlated extraction step at cuts crossed once by every path finds a solution at the cost of a constant number of evaluations.

cs.DS↗

Nanoscopic Multiplexing Optical Data Storage via Chip Fabrication

The accelerating growth of global data demands storage technologies combining high capacity, long lifetime and low energy consumption. Optical storage is attractive for archival applications, but conventional optical writing is constrained by diffraction-limited resolution and scalability, while thermal accumulation limits high-speed writing. Here we shift optical data writing from a serial optical process to a fabrication-defined process, in which lithographic masks govern where material modification occurs. Using patterned resist masks and ion implantation, we create fluorescent information voxels with nanoscale control over millimeter-scale areas. The platform supports multilevel and multidimensional encoding by independently encoding information in fluorescence intensity and wavelength. A machine-learning-assisted super-resolution reconstruction protocol is developed to recover information beyond conventional optical resolution for information readout. Overall, the framework reaches an areal storage capacity of 1.1 Tbit within the area of a standard 120-mm optical disc, approximately fivefold that of a conventional single-layer Blu-ray disc. This architecture establishes a micro/nanofabrication-compatible route to scalable optical memory and opens a pathway to wafer-scale archival storage.

physics.optics↗

Instance-Optimality of Bidirectional PageRank Estimation

We study the problem of estimating a vertex's PageRank within a constant relative error, with constant probability. We prove that an adaptive variant of the simple classic bidirectional algorithm is instance-optimal up to a polylogarithmic factor for all directed graphs of order $n$ whose maximum in- and out-degrees are at most a constant fraction of $n$. In other words, there is no correct algorithm that can be faster than our algorithm on any such graph by more than a polylogarithmic factor. We further extend the instance-optimality to all graphs in which at most a polylogarithmic number of vertices have unbounded degrees. This covers all sparse graphs with $\tilde{O}(n)$ edges. In addition, we provide a counterexample showing that the bidirectional algorithm is not instance-optimal for graphs whose degrees are mostly equal to $n$. We also consider weighted graphs and multigraphs. We show that the bidirectional algorithm is instance-optimal on \emph{all} multigraphs, but for weighted simple graphs, we have almost the same limitations as for unweighted simple graphs.

cs.DS↗

Instance-Optimality of Bidirectional Dijkstra on Simple Graphs

We study the shortest-path problem on graphs with positive real-valued edge weights. Given a source vertex $s$ and a target vertex $t$, the goal is to calculate the length of the shortest path from $s$ to $t$. We are particularly interested in instances that can be solved in sublinear time. Recently, Haeupler, Hladík, Rozhoň, Tarjan, and Tětek proved that (a version of) the bidirectional Dijkstra's algorithm is instance-optimal on positively weighted multigraphs, both directed and undirected, considering the number of vertices and edges queried by the algorithm. However, multigraphs are not the canonical setting for the shortest-path problem. The problem is typically formulated on simple graphs without loops and parallel edges. They therefore left as an open problem whether bidirectional Dijkstra remains instance-optimal on simple weighted graphs. We answer this question, but for simple graphs, the answer is more complex, depending on the setting. We show that bidirectional Dijkstra is still instance-optimal on simple undirected weighted graphs under the order-oblivious model, where incident edges are given in a random order. In contrast, under the order-dependent model, where incident edges have a given order, we show that bidirectional Dijkstra is not instance-optimal. For simple directed weighted graphs, we show that bidirectional Dijkstra is not instance-optimal under either the order-oblivious or the order-dependent model, being off by a factor of $Θ(m/n)$ in both cases. We further show that no algorithm can have instance-optimality ratio $o(m/n)$ under the order-dependent model, or under the order-oblivious model when $m=O(n\sqrt{n})$. On the positive side, the above results imply that bidirectional Dijkstra is instance-optimal up to logarithmic factors on all sparse directed and undirected graphs satisfying $m/n=\log^{O(1)} n$.

cs.DS↗

$\ell_p$-Norm Maximization over Zonotopes Is W[1]-Hard

We study $\ell_p$-norm maximization over zonotopes given by rational generators, with input length $L$. For fixed $p=a/b>1$, the exact Turing baseline runs in $n^{O(d)}b^{O(d)}\mathrm{poly}(L)$ time, but fixed-parameter tractability in the ambient dimension $d$ was open [FGHS25]. We prove W[1]-hardness and, under the Exponential Time Hypothesis (ETH), exclude $ρ_p(d)L^{o(d)}$ time, even for $5$-sparse generators, by encoding binary CSP constraints with normalized positive cap generators. We also give a deterministic $(1-\varepsilon)$-approximation with $\varepsilon^{-(d-1)/2}$ dependence and, among algorithms with fixed-degree polynomial dependence on $L$, rule out $(1/\varepsilon)^{o(d)}$ dependence under ETH. Support-function duality transfers the results to positive-output two-layer ReLU networks.

cs.CC↗

Approximate Graph Propagation Revisited: Dynamic Parameterized Queries, Tighter Bounds and Dynamic Updates

We revisit Approximate Graph Propagation (AGP), a unified framework which captures various graph propagation tasks, such as PageRank, feature propagation in Graph Neural Networks (GNNs), and graph-based Retrieval-Augmented Generation (RAG). Our work focuses on the settings of dynamic graphs and dynamic parameterized queries, where the underlying graphs evolve over time (updated by edge insertions or deletions) and the input query parameters are specified on the fly to fit application needs. Our first contribution is an interesting observation that the SOTA solution, AGP-Static, can be adapted to support dynamic parameterized queries; however several challenges remain unresolved. Firstly, the query time complexity of AGP-Static is based on an assumption of using an optimal algorithm for subset sampling in its query algorithm. Unfortunately, back to that time, such an algorithm did not exist; without such an optimal algorithm, an extra $O(\log^2 n)$ factor is required in the query complexity, where $n$ is the number of vertices in the graphs. Secondly, AGP-Static performs poorly on dynamic graphs, taking $O(n\log n)$ time to process each update. To address these challenges, we propose a new algorithm, AGP-Static++, which is simpler yet reduces roughly a factor of $O(\log^2 n)$ in the query complexity while preserving the approximation guarantees of AGP-Static. However, AGP-Static++ still requires $O(n)$ time to process each update. To better support dynamic graphs, we further propose AGP-Dynamic, which achieves $O(1)$ amortized time per update, significantly improving the aforementioned $O(n)$ per-update bound, while still preserving the query complexity and approximation guarantees. Last, our comprehensive experiments validate the theoretical improvements: compared to the baselines, our algorithm achieves speedups of up to $177\times$ on update time and $10\times$ on query efficiency.

cs.DS↗

PageRank Centrality in Directed Graphs with Bounded In-Degree

We study the computational complexity of locally estimating a node's PageRank centrality in a directed graph $G$. For any node $t$, its PageRank centrality $π(t)$ is defined as the probability that a random walk in $G$, starting from a uniformly chosen node, terminates at $t$, where each step terminates with a constant probability $α\in(0,1)$. To obtain a multiplicative $\big(1\pm O(1)\big)$-approximation of $π(t)$ with probability $Ω(1)$, the previously best upper bound is $O(n^{1/2}\min\{ Δ_{in}^{1/2},Δ_{out}^{1/2},m^{1/4}\})$ from [Wang, Wei, Wen, Yang, STOC '24], where $n$ and $m$ denote the number of nodes and edges in $G$, and $Δ_{in}$ and $Δ_{out}$ upper bound the in-degrees and out-degrees of $G$, respectively. Using a refinement of the proof in the same paper, we establish a lower bound of $Ω(n^{1/2}\min\{Δ_{in}^{1/2}/n^γ,Δ_{out}^{1/2}/n^γ,m^{1/4}\})$, where $γ=\frac{1}{2}(2\max\{\log_{1/(1-α)}Δ_{in},1\}-1)^{-1}$. As $γ$ only depends on $Δ_{in}$ and $n^γ=O(1)$ for $Δ_{in}=Ω\left(n^{Ω(1)}\right)$, the known upper bound is tight if we only parameterize the complexity by $n$, $m$, and $Δ_{out}$. However, there remains a gap of $Ω(n^γ)$ when considering $Δ_{in}$, and this gap is large when $Δ_{in}$ is small. In the extreme case where $Δ_{in}\le1/(1-α)$, we have $γ=1/2$, leading to a gap of $Ω(n^{1/2})$ between the bounds $O(n^{1/2})$ and $Ω(1)$. In this paper, we present a new algorithm that achieves the above lower bound (up to logarithmic factors). The algorithm assumes that $n$ and the bounds $Δ_{in}$ and $Δ_{out}$ are known in advance. Our key technique is a novel randomized backwards propagation process that only propagates selectively based on Monte Carlo estimated PageRank scores.

cs.DS↗

Mixing Time Matters: Accelerating Effective Resistance Estimation via Bidirectional Method

We study the problem of efficiently approximating the \textit{effective resistance} (ER) on undirected graphs, where ER is a widely used node proximity measure with applications in graph spectral sparsification, multi-class graph clustering, network robustness analysis, graph machine learning, and more. Specifically, given any nodes $s$ and $t$ in an undirected graph $G$, we aim to efficiently estimate the ER value $R(s,t)$ between nodes $s$ and $t$, ensuring a small absolute error $ε$. The previous best algorithm for this problem has a worst-case computational complexity of $\tilde{O}\left(\frac{L_{\max}^3}{ε^2 d^2}\right)$, where the value of $L_{\max}$ depends on the mixing time of random walks on $G$, $d = \min\{d(s), d(t)\}$, and $d(s)$, $d(t)$ denote the degrees of nodes $s$ and $t$, respectively. We improve this complexity to $\tilde{O}\left(\min\left\{\frac{L_{\max}^{7/3}}{ε^{2/3}}, \frac{L_{\max}^3}{ε^2d^2}, mL_{\max}\right\}\right)$, achieving a theoretical improvement of $\tilde{O}\left(\max\left\{\frac{L_{\max}^{2/3}}{ε^{4/3} d^2}, 1, \frac{L_{\max}^2}{ε^2 d^2 m}\right\}\right)$ over previous results. Here, $m$ denotes the number of edges. Given that $L_{\max}$ is often very large in real-world networks (e.g., $L_{\max} > 10^4$), our improvement on $L_{\max}$ is significant, especially for real-world networks. We also conduct extensive experiments on real-world and synthetic graph datasets to empirically demonstrate the superiority of our method. The experimental results show that our method achieves a $10\times$ to $1000\times$ speedup in running time while maintaining the same absolute error compared to baseline methods.

cs.SI↗

Measuring Fairness in Financial Transaction Machine Learning Models

Mastercard, a global leader in financial services, develops and deploys machine learning models aimed at optimizing card usage and preventing attrition through advanced predictive models. These models use aggregated and anonymized card usage patterns, including cross-border transactions and industry-specific spending, to tailor bank offerings and maximize revenue opportunities. Mastercard has established an AI Governance program, based on its Data and Tech Responsibility Principles, to evaluate any built and bought AI for efficacy, fairness, and transparency. As part of this effort, Mastercard has sought expertise from the Turing Institute through a Data Study Group to better assess fairness in more complex AI/ML models. The Data Study Group challenge lies in defining, measuring, and mitigating fairness in these predictions, which can be complex due to the various interpretations of fairness, gaps in the research literature, and ML-operations challenges.

cs.LG↗

Optimal Dynamic Parameterized Subset Sampling

In this paper, we study the Dynamic Parameterized Subset Sampling (DPSS) problem in the Word RAM model. In DPSS, the input is a set,~$S$, of~$n$ items, where each item,~$x$, has a non-negative integer weight,~$w(x)$. Given a pair of query parameters, $(α, β)$, each of which is a non-negative rational number, a parameterized subset sampling query on~$S$ seeks to return a subset $T \subseteq S$ such that each item $x \in S$ is selected in~$T$, independently, with probability $p_x(α, β) = \min \left\{\frac{w(x)}{α\sum_{x\in S} w(x)+β}, 1 \right\}$. More specifically, the DPSS problem is defined in a dynamic setting, where the item set,~$S$, can be updated with insertions of new items or deletions of existing items. Our first main result is an optimal algorithm for solving the DPSS problem, which achieves~$O(n)$ pre-processing time, $O(1+μ_S(α,β))$ expected time for each query parameterized by $(α, β)$, given on-the-fly, and $O(1)$ time for each update; here, $μ_S(α,β)$ is the expected size of the query result. At all times, the worst-case space consumption of our algorithm is linear in the current number of items in~$S$. Our second main contribution is a hardness result for the DPSS problem when the item weights are~$O(1)$-word float numbers, rather than integers. Specifically, we reduce Integer Sorting to the deletion-only DPSS problem with float item weights. Our reduction implies that an optimal algorithm for deletion-only DPSS with float item weights (achieving all the same bounds as aforementioned) implies an optimal algorithm for Integer Sorting. The latter remains an important open problem. Last but not least, a key technical ingredient for our first main result is an efficient algorithm for generating Truncated Geometric random variates in $O(1)$ expected time in the Word RAM model.

cs.DS↗

Revisiting Local PageRank Estimation on Undirected Graphs: Simple and Optimal

We propose a simple and optimal algorithm, BackMC, for local PageRank estimation in undirected graphs: given an arbitrary target node $t$ in an undirected graph $G$ comprising $n$ nodes and $m$ edges, BackMC accurately estimates the PageRank score of node $t$ while assuring a small relative error and a high success probability. The worst-case computational complexity of BackMC is upper bounded by $O\left(\frac{1}{d_{\mathrm{min}}}\cdot \min\left(d_t, m^{1/2}\right)\right)$, where $d_{\mathrm{min}}$ denotes the minimum degree of $G$, and $d_t$ denotes the degree of $t$, respectively. Compared to the previously best upper bound of $ O\left(\log{n}\cdot \min\left(d_t, m^{1/2}\right)\right)$ (VLDB '23), which is derived from a significantly more complex algorithm and analysis, our BackMC improves the computational complexity for this problem by a factor of $Θ\left(\frac{\log{n}}{d_{\mathrm{min}}}\right)$ with a much simpler algorithm. Furthermore, we establish a matching lower bound of $Ω\left(\frac{1}{d_{\mathrm{min}}}\cdot \min\left(d_t, m^{1/2}\right)\right)$ for any algorithm that attempts to solve the problem of local PageRank estimation, demonstrating the theoretical optimality of our BackMC. We conduct extensive experiments on various large-scale real-world and synthetic graphs, where BackMC consistently shows superior performance.

cs.DS↗

Revisiting Local Computation of PageRank: Simple and Optimal

We revisit the classic local graph exploration algorithm ApproxContributions proposed by Andersen, Borgs, Chayes, Hopcroft, Mirrokni, and Teng (WAW '07, Internet Math. '08) for computing an $ε$-approximation of the PageRank contribution vector for a target node $t$ on a graph with $n$ nodes and $m$ edges. We give a worst-case complexity bound of ApproxContributions as $O(nπ(t)/ε\cdot\min(Δ_{in},Δ_{out},\sqrt{m}))$, where $π(t)$ is the PageRank score of $t$, and $Δ_{in}$ and $Δ_{out}$ are the maximum in-degree and out-degree of the graph, resp. We also give a lower bound of $Ω(\min(Δ_{in}/δ,Δ_{out}/δ,\sqrt{m}/δ,m))$ for detecting the $δ$-contributing set of $t$, showing that the simple ApproxContributions algorithm is already optimal. We also investigate the computational complexity of locally estimating a node's PageRank centrality. We improve the best-known upper bound of $\widetilde{O}(n^{2/3}\cdot\min(Δ_{out}^{1/3},m^{1/6}))$ given by Bressan, Peserico, and Pretto (SICOMP '23) to $O(n^{1/2}\cdot\min(Δ_{in}^{1/2},Δ_{out}^{1/2},m^{1/4}))$ by simply combining ApproxContributions with the Monte Carlo simulation method. We also improve their lower bound of $Ω(\min(n^{1/2}Δ_{out}^{1/2},n^{1/3}m^{1/3}))$ to $Ω(n^{1/2}\cdot\min(Δ_{in}^{1/2},Δ_{out}^{1/2},m^{1/4}))$ if $\min(Δ_{in},Δ_{out})=Ω(n^{1/3})$, and to $Ω(n^{1/2-γ}(\min(Δ_{in},Δ_{out}))^{1/2+γ})$ if $\min(Δ_{in},Δ_{out})=o(n^{1/3})$, where $γ>0$ is an arbitrarily small constant. Our matching upper and lower bounds resolve the open problem of whether one can tighten the bounds given by Bressan, Peserico, and Pretto (FOCS '18, SICOMP '23). Remarkably, the techniques and analyses for proving all our results are surprisingly simple.

cs.DS↗

Efficient Algorithms for Personalized PageRank Computation: A Survey

Personalized PageRank (PPR) is a traditional measure for node proximity on large graphs. For a pair of nodes $s$ and $t$, the PPR value $π_s(t)$ equals the probability that an $α$-discounted random walk from $s$ terminates at $t$ and reflects the importance between $s$ and $t$ in a bidirectional way. As a generalization of Google's celebrated PageRank centrality, PPR has been extensively studied and has found multifaceted applications in many fields, such as network analysis, graph mining, and graph machine learning. Despite numerous studies devoted to PPR over the decades, efficient computation of PPR remains a challenging problem, and there is a dearth of systematic summaries and comparisons of existing algorithms. In this paper, we recap several frequently used techniques for PPR computation and conduct a comprehensive survey of various recent PPR algorithms from an algorithmic perspective. We classify these approaches based on the types of queries they address and review their methodologies and contributions. We also discuss some representative algorithms for computing PPR on dynamic graphs and in parallel or distributed environments.

cs.DS↗

Optimal Dynamic Subset Sampling: Theory and Applications

We study the fundamental problem of sampling independent events, called subset sampling. Specifically, consider a set of $n$ events $S=\{x_1, \ldots, x_n\}$, where each event $x_i$ has an associated probability $p(x_i)$. The subset sampling problem aims to sample a subset $T \subseteq S$, such that every $x_i$ is independently included in $S$ with probability $p_i$. A naive solution is to flip a coin for each event, which takes $O(n)$ time. However, the specific goal is to develop data structures that allow drawing a sample in time proportional to the expected output size $μ=\sum_{i=1}^n p(x_i)$, which can be significantly smaller than $n$ in many applications. The subset sampling problem serves as an important building block in many tasks and has been the subject of various research for more than a decade. However, most of the existing subset sampling approaches are conducted in a static setting, where the events or their associated probability in set $S$ is not allowed to be changed over time. These algorithms incur either large query time or update time in a dynamic setting despite the ubiquitous time-evolving events with changing probability in real life. Therefore, it is a pressing need, but still, an open problem, to design efficient dynamic subset sampling algorithms. In this paper, we propose ODSS, the first optimal dynamic subset sampling algorithm. The expected query time and update time of ODSS are both optimal, matching the lower bounds of the subset sampling problem. We present a nontrivial theoretical analysis to demonstrate the superiority of ODSS. We also conduct comprehensive experiments to empirically evaluate the performance of ODSS. Moreover, we apply ODSS to a concrete application: influence maximization. We empirically show that our ODSS can improve the complexities of existing influence maximization algorithms on large real-world evolving social networks.

cs.DS↗

An Effective Two-stage Training Paradigm Detector for Small Dataset

Learning from the limited amount of labeled data to the pre-train model has always been viewed as a challenging task. In this report, an effective and robust solution, the two-stage training paradigm YOLOv8 detector (TP-YOLOv8), is designed for the object detection track in VIPriors Challenge 2023. First, the backbone of YOLOv8 is pre-trained as the encoder using the masked image modeling technique. Then the detector is fine-tuned with elaborate augmentations. During the test stage, test-time augmentation (TTA) is used to enhance each model, and weighted box fusion (WBF) is implemented to further boost the performance. With the well-designed structure, our approach has achieved 30.4% average precision from 0.50 to 0.95 on the DelftBikes test set, ranking 4th on the leaderboard.

cs.CV↗