United States

Neural networks require careful weight initialization to prevent signals from exploding or vanishing.  Existing initialization schemes solve this problem in specific cases by assuming that the network has a certain activation function or topology.  It is difficult to derive such weight initialization strategies, and modern architectures therefore often use these same initialization schemes even though their assumptions do not hold. This paper introduces AutoInit, a weight initialization algorithm that automatically adapts to different neural network architectures.  By analytically tracking the mean and variance of signals as they propagate through the network, AutoInit appropriately scales the weights at each layer to avoid exploding or vanishing signals.  Experiments demonstrate that AutoInit improves performance of convolutional, residual, and transformer networks across a range of activation function, dropout, weight decay, learning rate, and normalizer settings, and does so more reliably than data-dependent initialization methods.  This flexibility allows AutoInit to initialize models for everything from small tabular tasks to large datasets such as ImageNet.  Such generality turns out particularly useful in neural architecture search and in activation function discovery.  In these settings, AutoInit initializes each candidate appropriately, making performance evaluations more accurate. AutoInit thus serves as an automatic configuration tool that makes design of new neural network architectures more robust. The AutoInit package provides a wrapper around TensorFlow models and is available at https://github.com/cognizant-ai-labs/autoinit.

AAAI 2023

AutoInit: Analytic Signal-Preserving Weight Initialization for Neural Networks

analytic

weight initialization

signal

automatic

## Welcome to EMNLP 2024! 
We are excited to welcome you to one of the most prominent conferences in the field of Natural Language Processing. This year, EMNLP 2024 is being held in a hybrid format,
offering both virtual and in-person participation in beautiful Miami. Due to a record-breaking number of submissions, we've expanded the total number of accepted papers to accommodate more cutting-edge research from around the globe.
### [Conference Handbook](https://drive.google.com/file/d/1WPROgxjLAC96AJL7Ugy0tEnYm7dkrbHt/view?usp=sharing)

You are required to register for this event. **Please register [here](https://2024.emnlp.org/registration/).** The EMNLP 2024 event page on Underline will be open to public one week prior to the event.

Please register!

EMNLP 2024

EMNLP 2024 will take place in Miami, Florida from Nov 12th to Nov 16th, 2024, at the Hyatt Regency Miami Hote and on Underline for remote participants.

poster

The purpose of the AAAI conference series is to promote research in Artificial Intelligence (AI) and foster scientific exchange between researchers, practitioners, scientists, students, and engineers across the entirety of AI and its affiliated disciplines. AAAI-23 is the Thirty-Seventh AAAI Conference on Artificial Intelligence. The theme of this conference is to create collaborative bridges within and beyond AI. Like previous AAAI conferences, AAAI-23 will feature technical paper presentations, special tracks, invited speakers, workshops, tutorials, poster sessions, senior member presentations, competitions, and exhibit programs, and two new activities: a Bridge Program and a Lab Program. Many of these activities are tailored to the theme of bridges and all are selected according to the highest standards, with additional programs for students and young researchers. 
AAAI is providing you with a conference planner, which you can use to help organize your itinerary of activities. This includes talks to attend in person, talks to attend remotely, breaks with colleagues and your site seeing activities. To access this conference planner, please go to [https://aaai-2023.takemobi.io](https://aaai-2023.takemobi.io).

In order to access this site, you need to register. If you haven't already, please register [here](https://aaai.org/Conferences/AAAI-23/registration/).


The purpose of the AAAI conference series is to promote research in Artificial Intelligence (AI) and foster scientific exchange between researchers, practitioners, scientists, students, and engineers across the entirety of AI and its affiliated disciplines.

 With the development of advanced driver assistance systems~(ADAS) and autonomous vehicles, conducting experiments in various scenarios becomes an urgent need. Although having been capable of synthesizing photo-realistic street scenes, conventional image-to-image translation methods cannot produce coherent scenes due to the lack of 3D information. In this paper, a large-scale neural rendering method is proposed to synthesize the autonomous driving scene~(READ), which makes it possible to generate large-scale driving scenes in real time on a PC through a variety of sampling schemes. In order to effectively represent driving scenarios, we propose an ω rendering network to learn neural descriptors from sparse point clouds. Our model can not only synthesize photo-realistic driving scenes but also stitch and edit them. The promising experimental results show that our model performs well in large-scale driving scenarios.


READ: Large-Scale Neural Scene Rendering for Autonomous Driving

Submodular maximization has attracted extensive attention due to its numerous applications in machine learning and artificial intelligence. Many real-world problems require maximizing multiple submodular objective functions at the same time. In such cases, a common approach is to select a representative subset of Pareto optimal solutions with different trade-offs among multiple objectives. To this end, in this paper, we investigate the regret ratio minimization (RRM) problem in multi-objective submodular maximization, which aims to find at most k solutions to best approximate all Pareto optimal solutions w.r.t. any linear combination of objective functions. We propose a novel HS-RRM algorithm by transforming RRM into HittingSet problems based on the notions of $\epsilon$-kernel and $\delta$-net, where any $\alpha$-approximation algorithm for single-objective submodular maximization is used as an oracle. We prove that the maximum regret ratio (MRR) of the output of HS-RRM is bounded by $1 - \alpha + O\big( (k-d)^{-\frac{2}{d-1}} \big)$, where d is the number of objectives, which improves upon the previous best-known bound of $1 - \alpha + O\big( (k-d)^{-\frac{1}{d-1}} \big)$ and is nearly asymptotically optimal for any fixed d. Experiments on real-world and synthetic data confirm that HS-RRM achieves lower MRRs than existing algorithms.

Improved Algorithm for Regret Ratio Minimization in Multi-Objective Submodular Maximization

Privacy in AI remains a topic that draws attention from researchers and the general public in recent years. As one way to implement privacy-preserving AI, differentially private learning is a framework that enables AI models to use differential privacy (DP). To achieve DP in the learning process, existing algorithms typically limit the magnitude of gradients with a constant clipping, which requires carefully tuned due to its significant impact on model performance. As a solution to this issue, latest works NSGD and Auto-S innovatively propose to use normalization instead of clipping to avoid hyperparameter tuning. However, normalization-based approaches like NSGD and Auto-S rely on a monotonic weight function, which imposes excessive weight on small gradient samples and introduces extra deviation to the update. In this paper, we propose a Differentially Private Per-Sample Adaptive Clipping (DP-PSAC) algorithm based on a non-monotonic adaptive weight function, which guarantees privacy without the typical hyperparameter tuning process of using a constant clipping while significantly reducing the deviation between the update and true batch-averaged gradient. We provide a rigorous theoretical convergence analysis and show that with convergence rate at the same order, the proposed algorithm achieves a lower non-vanishing bound, which is maintained over training iterations, compared with NSGD/Auto-S.  In addition, through extensive experimental evaluation, we show that DP-PSAC outperforms or matches the state-of-the-art methods on multiple main-stream vision and language tasks. 

Differentially Private Learning with Per-Sample Adaptive Clipping

Jigsaw puzzle solving has recently become an emerging research area. The developed techniques have been widely used in applications beyond puzzle solving. This paper focuses on solving Jigsaw Puzzles with Large Eroded Gaps (JPwLEG). We formulate the puzzle reassembly as a combinatorial optimization problem and propose a Siamese-Discriminant Deep Reinforcement Learning (SD$^2$RL) to solve it. A Deep Q-network (DQN) is designed to visually understand the puzzles, which consists of two sets of Siamese Discriminant Networks, one set to perceive the pairwise relations between vertical neighbors and another set for horizontal neighbors. The proposed DQN considers not only the evidence from the incumbent fragment but also the support from its four neighbors. The DQN is trained using replay experience with carefully designed rewards to guide the search for a sequence of fragment swaps to reach the correct puzzle solution. Two JPwLEG datasets are constructed to evaluate the proposed method, and the experimental results show that the proposed SD$^2$RL significantly outperforms state-of-the-art methods.

Siamese-Discriminant Deep Reinforcement Learning for Solving Jigsaw Puzzles with Large Eroded Gaps

Constructing useful representations across a large number of tasks is a key requirement for sample-efficient intelligent systems. A traditional idea in multitask learning (MTL) is building a shared representation across tasks which can then be adapted to new tasks by tuning last layers. A desirable refinement of using a shared one-fits-all representation is to construct task-specific representations. To this end, recent PathNet/muNet architectures represent individual tasks as pathways within a larger supernet. The subnetworks induced by pathways can be viewed as task-specific representations that are composition of modules within supernet's computation graph. This work explores the pathways proposal from the lens of statistical learning: We first develop novel generalization bounds for empirical risk minimization problems learning multiple tasks over multiple paths (Multipath MTL). In conjunction, we formalize the benefits of resulting multipath representation when adapting to new downstream tasks. Our bounds are expressed in terms of Gaussian complexity, lead to tangible guarantees for the class of linear representations, and provide novel insights into the quality and benefits of a multipath representation. When computation graph is a tree, Multipath MTL hierarchically clusters the tasks and builds cluster-specific representations. We provide further discussion and experiments for hierarchical MTL and rigorously identify the conditions under which Multipath MTL is provably superior to traditional MTL approaches with shallow supernets.

Provable Pathways: Learning Multiple Tasks over Multiple Paths

Scene Graph Generation (SGG) aims to capture the semantic information in an image and build a structured representation, which facilitates downstream tasks. The current challenge in SGG is to tackle the biased predictions caused by the long-tailed distribution of predicates. Since multiple predicates in SGG are coupled in an image, existing data re-balancing methods cannot completely balance the head and tail predicates. In this work, a decoupled learning framework is proposed for unbiased scene graph generation by using attribute-guided predicate features to construct a balanced training set. Specifically, the predicate recognition is decoupled into Predicate Feature Representation Learning (PFRL) and predicate classifier training with a class-balanced predicate feature set, which is constructed by our proposed Attribute-guided Predicate Feature Generation (A-PFG) model. In the A-PFG model, we first define the class labels of <subject-predicate-object> and corresponding visual feature as attributes to describe a predicate. Then the predicate feature and the attribute embedding are mapped into a shared hidden space by a dual Variational Auto-encoder (VAE), and finally the synthetic predicate features are forced to learn the contextual information in the attributes via cross reconstruction and distribution alignment. To demonstrate the effectiveness of our proposed method, our decoupled learning framework and A-PFG model are applied to various SGG models. The empirical results show that our method is substantially improved on all benchmarks and achieves new state-of-the-art performance for unbiased scene graph generation. Our code is available at https://github.com/wanglei0618/A-PFG.

Learning to Generate an Unbiased Scene Graph by Using Attribute-guided Predicate Features

In classical planning, the aim is to find a sequence of deterministic actions leading from the initial to a goal state. In this work, we consider the scenario where a party who knows the solution to a planning task, called the prover, wants to convince a second party, the verifier, that it has the solution without revealing any information about the solution itself. This is relevant in domains where privacy is important, for example when plans contain sensitive information or when the solution should not be revealed upfront.  We achieve this by introducing a zero-knowledge protocol for plan existence. By restricting ourselves to tasks with polynomially-bounded plan length, we are able to construct a protocol that can be run efficiently by both prover and verifier. The resulting protocol does not rely on any reduction, has a constant number of rounds, and runs in time polynomial in the size of the task.

Zero-Knowledge Proofs for Classical Planning Problems

Named Entity Recognition (NER) models capable of Continual Learning (CL) are realistically valuable in areas where entity types continuously increase (e.g., personal assistants). Meanwhile the learning paradigm of NER advances to new patterns such as the span-based methods. However, its potential to CL has not been fully explored. In this paper, we propose SpanKL, a simple yet effective Span-based model with Knowledge distillation (KD) to preserve memories and multi-Label prediction to prevent conflicts in CL-NER. Unlike prior sequence labeling approaches, the inherently independent modeling in span and entity level with the designed coherent optimization on SpanKL promotes its learning at
each incremental step and mitigates the forgetting. Experiments on synthetic CL datasets derived from OntoNotes and Few-NERD show that SpanKL significantly outperforms previous SoTA in many aspects, and obtains the smallest gap
from CL to the upper bound revealing its high practiced value.

A Neural Span-Based Continual Named Entity Recognition Model

Recently, fake news forgery technology has become more and more sophisticated, and even the profiles of participants may be faked, which challenges the robustness and effectiveness of traditional detection methods involving text or user identity. Most propagation-only approaches mainly rely on neural networks to learn the diffusion pattern of individual news, which is insufficient to describe the differences in news spread ability, and also ignores the valuable global connections of news and users, limiting the performance of detection. Therefore, we propose a joint learning model named HG-SL, which is blind to news content and user identities, but capable of catching the differences between true and fake news in the early stages of propagation through global and local user spreading behavior. Specifically, we innovatively design a Hypergraph-based Global interaction learning module to capture the global preferences of users from their co-spreading relationships, and introduce node centrality encoding to complement user influence in hypergraph learning. Moreover, the designed Self-attention-based Local context learning module first introduce spread status to highlight the propagation ability of news and users, thus providing additional signals for verifying news authenticity. Experiments on real-world datasets indicate that our HG-SL, which solely relies on user behavior, outperforms SOTA baselines utilizing multidimensional features in both fake news detection and early detection task.

HG-SL: Jointly Learning of Global and Local User Spreading Behavior for Fake News Early Detection

The actor-critic (AC) reinforcement learning algorithms have been the powerhouse behind many challenging applications. Nevertheless, its convergence is fragile in general. To study its instability, existing works mostly consider the uncommon double-loop variant or basic models with finite state and action space. We investigate the more practical single-sample two-timescale AC for solving the canonical linear quadratic regulator (LQR) problem, where the actor and the critic update only once with a single sample in each iteration on an unbounded continuous state and action space. Existing analysis cannot conclude the convergence for such a challenging case. We develop a new analysis framework that allows establishing the global convergence to an $\epsilon$-optimal solution with at most an $\tilde{\mathcal{O}}(\epsilon^{-2.5})$ sample complexity. To our knowledge, this is the first finite-time convergence analysis for the single sample two-timescale AC for solving LQR with global optimality. The sample complexity improves those of other variants by orders, which sheds light on the practical wisdom of single sample algorithms. We also further validate our theoretical findings via comprehensive simulation comparisons.

Downloads

Next from AAAI 2023

READ: Large-Scale Neural Scene Rendering for Autonomous Driving

Similar lecture

An Empirical Comparison of Vocabulary Expansion and Initialization Approaches For Language Models