Traditional methods of computing WAR (wins above replacement) for pitchers are based on an invalid mathematical foundation. Consequently, these metrics, which produce reasonable values for many pitchers, can be substantially inaccurate for some. Specifically, FanGraphs and Baseball Reference compute a pitcher's WAR as a function of his performance averaged over the entire season. This is wrong because WAR must be a convex function of the number of runs allowed by the pitcher in a game. Hence we propose a new way to compute WAR for starting pitchers: Grid WAR (GWAR). The idea is to compute a starter's GWAR for each of his individual games, and define a starter's seasonal GWAR as the sum of the GWAR of each of his games. We show that GWAR is indeed a convex function in the number of runs allowed during a game. As such, GWAR accounts for a fundamental baseball principle that not all runs allowed have the same impact in determining the outcome of a game: for instance, the difference in GWAR between allowing 1 run in a game instead of 0 is much greater than the difference in GWAR between allowing 6 runs in a game instead of 5. Moreover, Jensen's inequality implies that, by ignoring the convexity of WAR, current implementations of WAR undervalue certain pitchers, particularly those who allow few runs (specifically, 0 or 1 run) in many games. It also unfairly penalizes pitchers who are credited with a large number of runs in a short outing. These flaws are corrected by GWAR.
相關內容
Tailor-made for massive connectivity and sporadic access, grant-free random access has become a promising candidate access protocol for massive machine-type communications (mMTC). Compared with conventional grant-based protocols, grant-free random access skips the exchange of scheduling information to reduce the signaling overhead, and facilitates sharing of access resources to enhance access efficiency. However, some challenges remain to be addressed in the receiver design, such as unknown identity of active users and multi-user interference (MUI) on shared access resources. In this work, we deal with the problem of joint user activity and data detection for grant-free random access. Specifically, the approximate message passing (AMP) algorithm is first employed to mitigate MUI and decouple the signals of different users. Then, we extend the data symbol alphabet to incorporate the null symbols from inactive users. In this way, the joint user activity and data detection problem is formulated as a clustering problem under the Gaussian mixture model. Furthermore, in conjunction with the AMP algorithm, a variational Bayesian inference based clustering (VBIC) algorithm is developed to solve this clustering problem. Simulation results show that, compared with state-of-art solutions, the proposed AMP-combined VBIC (AMP-VBIC) algorithm achieves a significant performance gain in detection accuracy.
A wide range of NLP tasks benefit from the fine-tuning of pretrained language models (PLMs). However, a number of redundant parameters which contribute less to the downstream task are observed in a directly fine-tuned model. We consider the gap between pretraining and downstream tasks hinders the training of these redundant parameters, and results in a suboptimal performance of the overall model. In this paper, we present PATS (Perturbation According To Sensitivity), a noisy training mechanism which considers each parameter's importance in the downstream task to help fine-tune PLMs. The main idea of PATS is to add bigger noise to parameters with lower sensitivity and vice versa, in order to activate more parameters' contributions to downstream tasks without affecting the sensitive ones much. Extensive experiments conducted on different tasks of the GLUE benchmark show PATS can consistently empower the fine-tuning of different sizes of PLMs, and the parameters in the well-performing models always have more concentrated distributions of sensitivities, which experimentally proves the effectiveness of our method.
We develop a general method to study the Fisher information distance in central limit theorem for nonlinear statistics. We first construct completely new representations for the score function. We then use these representations to derive quantitative estimates for the Fisher information distance. To illustrate the applicability of our approach, explicit rates of Fisher information convergence for quadratic forms and the functions of sample means are provided. For the sums of independent random variables, we obtain the Fisher information bounds without requiring the finiteness of Poincar\'e constant. Our method can also be used to bound the Fisher information distance in non-central limit theorems.
Sampling from a target measure whose density is only known up to a normalization constant is a fundamental problem in computational statistics and machine learning. In this paper, we present a new optimization-based method for sampling called mollified interaction energy descent (MIED). MIED minimizes a new class of energies on probability measures called mollified interaction energies (MIEs). These energies rely on mollifier functions -- smooth approximations of the Dirac delta originated from PDE theory. We show that as the mollifier approaches the Dirac delta, the MIE converges to the chi-square divergence with respect to the target measure and the gradient flow of the MIE agrees with that of the chi-square divergence. Optimizing this energy with proper discretization yields a practical first-order particle-based algorithm for sampling in both unconstrained and constrained domains. We show experimentally that for unconstrained sampling problems our algorithm performs on par with existing particle-based algorithms like SVGD, while for constrained sampling problems our method readily incorporates constrained optimization techniques to handle more flexible constraints with strong performance compared to alternatives.
Real-world optimization problems may have a different underlying structure. In black-box optimization, the dependencies between decision variables remain unknown. However, some techniques can discover such interactions accurately. In Large Scale Global Optimization (LSGO), problems are high-dimensional. It was shown effective to decompose LSGO problems into subproblems and optimize them separately. The effectiveness of such approaches may be highly dependent on the accuracy of problem decomposition. Many state-of-the-art decomposition strategies are derived from Differential Grouping (DG). However, if a given problem consists of non-additively separable subproblems, DG-based strategies may discover many non-existing interactions. On the other hand, monotonicity checking strategies proposed so far do not report non-existing interactions for any separable subproblems but may miss discovering many of the existing ones. Therefore, we propose Incremental Recursive Ranking Grouping (IRRG) that suffers from none of these flaws. IRRG consumes more fitness function evaluations than the recent DG-based propositions, e.g., Recursive DG 3 (RDG3). Nevertheless, the effectiveness of the considered Cooperative Co-evolution frameworks after embedding IRRG or RDG3 was similar for problems with additively separable subproblems that are suitable for RDG3. After replacing the additive separability with non-additive, embedding IRRG leads to results of significantly higher quality.
We present a lazy incremental search algorithm, Lifelong-GLS (L-GLS), along with its bounded suboptimal version, Bounded L-GLS (B-LGLS) that combine the search efficiency of incremental search algorithms with the evaluation efficiency of lazy search algorithms for fast replanning in problem domains where edge-evaluations are more expensive than vertex-expansions. The proposed algorithms generalize Lifelong Planning A* (LPA*) and its bounded suboptimal version, Truncated LPA* (TLPA*), within the Generalized Lazy Search (GLS) framework, so as to restrict expensive edge evaluations only to the current shortest subpath when the cost-to-come inconsistencies are propagated during repair. We also present dynamic versions of the L-GLS and B-LGLS algorithms, called Generalized D* (GD*) and Bounded Generalized D* (B-GD*), respectively, for efficient replanning with non-stationary queries, designed specifically for navigation of mobile robots. We prove that the proposed algorithms are complete and correct in finding a solution that is guaranteed not to exceed the optimal solution cost by a user-chosen factor. Our numerical and experimental results support the claim that the proposed integration of the incremental and lazy search frameworks can help find solutions faster compared to the regular incremental or regular lazy search algorithms when the underlying graph representation changes often.
One of the most promising applications of the IoT is the Smart Grid (SG). Integrating SG's data communications network into the power grid allows gathering and analyzing information from power lines, distribution power stations, and end users. A smart grid (SG) requires a prompt and dependable connection to provide real-time monitoring through the IoT. Hence 5G could be considered a catalyst for upgrading the existing power grid systems. Nonetheless, the additional attack surface of information infrastructure has been brought about by the widespread adoption of ubiquitous connectivity in 5G, to which the typical information security system in the smart grid cannot respond promptly. Therefore, guaranteeing the Privacy and Security of a network in a threatening, ever-changing environment requires groundbreaking architectures that go well beyond the limitations of traditional, static security measures. With "Continuous Identity Authentication and Dynamic Access Control" as its foundation, this article analyzes the Zero Trust (ZT) architecture specific to the power system of IoT and uses that knowledge to develop a security protection architecture.
Action anticipation involves predicting future actions having observed the initial portion of a video. Typically, the observed video is processed as a whole to obtain a video-level representation of the ongoing activity in the video, which is then used for future prediction. We introduce ANTICIPATR which performs long-term action anticipation leveraging segment-level representations learned using individual segments from different activities, in addition to a video-level representation. We propose a two-stage learning approach to train a novel transformer-based model that uses these two types of representations to directly predict a set of future action instances over any given anticipation duration. Results on Breakfast, 50Salads, Epic-Kitchens-55, and EGTEA Gaze+ datasets demonstrate the effectiveness of our approach.
Bid optimization for online advertising from single advertiser's perspective has been thoroughly investigated in both academic research and industrial practice. However, existing work typically assume competitors do not change their bids, i.e., the wining price is fixed, leading to poor performance of the derived solution. Although a few studies use multi-agent reinforcement learning to set up a cooperative game, they still suffer the following drawbacks: (1) They fail to avoid collusion solutions where all the advertisers involved in an auction collude to bid an extremely low price on purpose. (2) Previous works cannot well handle the underlying complex bidding environment, leading to poor model convergence. This problem could be amplified when handling multiple objectives of advertisers which are practical demands but not considered by previous work. In this paper, we propose a novel multi-objective cooperative bid optimization formulation called Multi-Agent Cooperative bidding Games (MACG). MACG sets up a carefully designed multi-objective optimization framework where different objectives of advertisers are incorporated. A global objective to maximize the overall profit of all advertisements is added in order to encourage better cooperation and also to protect self-bidding advertisers. To avoid collusion, we also introduce an extra platform revenue constraint. We analyze the optimal functional form of the bidding formula theoretically and design a policy network accordingly to generate auction-level bids. Then we design an efficient multi-agent evolutionary strategy for model optimization. Offline experiments and online A/B tests conducted on the Taobao platform indicate both single advertiser's objective and global profit have been significantly improved compared to state-of-art methods.
Since hardware resources are limited, the objective of training deep learning models is typically to maximize accuracy subject to the time and memory constraints of training and inference. We study the impact of model size in this setting, focusing on Transformer models for NLP tasks that are limited by compute: self-supervised pretraining and high-resource machine translation. We first show that even though smaller Transformer models execute faster per iteration, wider and deeper models converge in significantly fewer steps. Moreover, this acceleration in convergence typically outpaces the additional computational overhead of using larger models. Therefore, the most compute-efficient training strategy is to counterintuitively train extremely large models but stop after a small number of iterations. This leads to an apparent trade-off between the training efficiency of large Transformer models and the inference efficiency of small Transformer models. However, we show that large models are more robust to compression techniques such as quantization and pruning than small models. Consequently, one can get the best of both worlds: heavily compressed, large models achieve higher accuracy than lightly compressed, small models.
小貼士
登錄享
相關主題
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191