VLDB125.DVI

Synthetic 1 98.9% 9.5% Synthetic 2 99.3% 8.9% 10 20 30 40 50 60 70 80 90 100 0 5 10 15 20 25 30 R E LA T IV E R U N N IN G T IM E DATA DIMENSIONALITY SCALABILITY WITH DATA DIMENSIONALITY Fig. 9. Scaling of [...] dimensionality [9] can be interpreted slightly differently to infer that each point is as good as an outlier in high-dimensional space. This is because if all pairs of points are almost equidistant [9], then [...] 6 7 8 9 10 x 10 5 0 200 400 600 800 1000 1200 R E LA T IV E R U N N IN G T IM E DATA SIZE SCALABILITY WITH DATA SIZE Fig. 7. Scaling of running time with database size (Syn. 1) 0 1 2 3 4 5 6 7 8 9 10 x …

leung_ossm.dvi

s * I , I 3 RC * 3 s I , I . Greedy . 9 s I , I 9 (a) Pure strategies with = 500 Hybrid strategy Segmentation time Speedup Random-RC * 3 * s .I 9 , I , Random-Greedy 3, 3 3 s I * , I 3 (b) [...] the following specific contributions:9 In Section 3, we present the OSSM and show how it pro- vides a light-weight mechanism to reduce the number of candidate patterns.9 In Section 4, we consider the segment [...] algorithms that do not rely on candidate generation (e.g., FP-growth [8], CHARM [21], and GenMax [20]). 9 Given that the OSSM created with /?0 ) segments will consume too much space, in Section 5 we con- sider …

rieck10a.dvi

APPROXIMATE TREE KERNELS 1 2 3 4 5 0.9 0.92 0.94 0.96 0.98 1 A re a un de r th e R O C c ur ve Selected symbols (N) ATK PTK (a) Classification performance 0 50 100 150 10 −9 10 −7 10 −5 10 −3 10 −1 10 1 10 [...] saturates for samples containing at least 120 571 RIECK, KRUEGER, BREFELD AND M ÜLLER 1 2 3 4 5 6 7 8 9 10 0.9 0.92 0.94 0.96 0.98 1 A re a un de r th e R O C c ur ve Selected symbols (N) ATK PTK (a) Classification [...] rieck10a.dvi Journal of Machine Learning Research 11 (2010) 555-580 Submitted 12/08; Revised 9/09; Published 2/10 Approximate Tree Kernels Konrad Rieck RIECK@CS.TU-BERLIN.DE Technische Universität Berlin …

construction ūk ∈ GSBV p(U ; Rn). By (4.9) and the fact that A3 k ⊂ A2 k we have Ln(A3 k) → 0 as k → +∞ from which (4.1) follows. On the other hand as Jūk ⊆ ∂A3 k (4.9) and (4.10) yield Fk(uk) (1 − η)βiHn−1(Jūk [...] that converges in measure to u ∈ L0(U ; Rn). Assume moreover that Fk(uk) C and that (2.9) holds. Then, for W as (2.9) lim inf k→+∞ Fk(uk, A) α ˆ A W (x, e(u)) dx ∀A ∈ A(U) , https://doi.org/10.1017/prm [...] energy only depends on f through its derivative at the origin. A similar effect has been observed in [9] where the analogue of (1.3) for energies depending on the full deformation gradient was taken into …

TU Dortmund -- Computer Science VIII

Systems (Vol. 29). Curran Associates, Inc. Available at https://papers.nips.cc/paper/2016/hash/9d2682367c3935defcb1f9e247a97c0d-Abstract.html Zafar, M.B., Valera, I., Rogriguez, M.G. & Gummadi, K.P.. (2017) [...] https://robustbench.github.io/ (guter Ausgangspunkt https://colab.research.google.com/drive/1MQY_7O9vj7ixD5ilVRbdQwlNPFvxifHV#scrollTo=ONmXzKqvAmWX ) https://github.com/bethgelab/foolbox   C. Fairness …