Ziming Liu

Yilun Xu, Ziming Liu, Max Tegmark et al. · mit

We propose a new "Poisson flow" generative model (PFGM) that maps a uniform distribution on a high-dimensional hemisphere into any data distribution. We interpret the data points as electrical charges on the $z=0$ hyperplane in a space augmented with an additional dimension $z$, generating a high-dimensional electric field (the gradient of the solution to Poisson equation). We prove that if these charges flow upward along electric field lines, their initial distribution in the $z=0$ plane transforms into a distribution on the hemisphere of radius $r$ that becomes uniform in the $r \to\infty$ limit. To learn the bijective transformation, we estimate the normalized field in the augmented space. For sampling, we devise a backward ODE that is anchored by the physically meaningful additional dimension: the samples hit the unaugmented data manifold when the $z$ reaches zero. Experimentally, PFGM achieves current state-of-the-art performance among the normalizing flow models on CIFAR-10, with an Inception score of $9.68$ and a FID score of $2.35$. It also performs on par with the state-of-the-art SDE approaches while offering $10\times $ to $20 \times$ acceleration on image generation tasks. Additionally, PFGM appears more tolerant of estimation errors on a weaker network architecture and robust to the step size in the Euler method. The code is available at https://github.com/Newbeeer/poisson_flow .

Yilun Xu, Ziming Liu, Yonglong Tian et al. · mit

We introduce a new family of physics-inspired generative models termed PFGM++ that unifies diffusion models and Poisson Flow Generative Models (PFGM). These models realize generative trajectories for $N$ dimensional data by embedding paths in $N{+}D$ dimensional space while still controlling the progression with a simple scalar norm of the $D$ additional variables. The new models reduce to PFGM when $D{=}1$ and to diffusion models when $D{\to}\infty$. The flexibility of choosing $D$ allows us to trade off robustness against rigidity as increasing $D$ results in more concentrated coupling between the data and the additional variable norms. We dispense with the biased large batch field targets used in PFGM and instead provide an unbiased perturbation-based objective similar to diffusion models. To explore different choices of $D$, we provide a direct alignment method for transferring well-tuned hyperparameters from diffusion models ($D{\to} \infty$) to any finite $D$ values. Our experiments show that models with finite $D$ can be superior to previous state-of-the-art diffusion models on CIFAR-10/FFHQ $64{\times}64$ datasets, with FID scores of $1.91/2.43$ when $D{=}2048/128$. In class-conditional setting, $D{=}2048$ yields current state-of-the-art FID of $1.74$ on CIFAR-10. In addition, we demonstrate that models with smaller $D$ exhibit improved robustness against modeling errors. Code is available at https://github.com/Newbeeer/pfgmpp

Yilun Xu, Mingyang Deng, Xiang Cheng et al. · mit

Generative processes that involve solving differential equations, such as diffusion models, frequently necessitate balancing speed and quality. ODE-based samplers are fast but plateau in performance while SDE-based samplers deliver higher sample quality at the cost of increased sampling time. We attribute this difference to sampling errors: ODE-samplers involve smaller discretization errors while stochasticity in SDE contracts accumulated errors. Based on these findings, we propose a novel sampling algorithm called Restart in order to better balance discretization errors and contraction. The sampling method alternates between adding substantial noise in additional forward steps and strictly following a backward ODE. Empirically, Restart sampler surpasses previous SDE and ODE samplers in both speed and accuracy. Restart not only outperforms the previous best SDE results, but also accelerates the sampling speed by 10-fold / 2-fold on CIFAR-10 / ImageNet $64 \times 64$. In addition, it attains significantly better sample quality than ODE samplers within comparable sampling times. Moreover, Restart better balances text-image alignment/visual quality versus diversity than previous samplers in the large-scale text-to-image Stable Diffusion model pre-trained on LAION $512 \times 512$. Code is available at https://github.com/Newbeeer/diffusion_restart_sampling

Ziqian Zhong, Ziming Liu, Max Tegmark et al. · microsoft-research, mit

Do neural networks, trained on well-understood algorithmic tasks, reliably rediscover known algorithms for solving those tasks? Several recent studies, on tasks ranging from group arithmetic to in-context linear regression, have suggested that the answer is yes. Using modular addition as a prototypical problem, we show that algorithm discovery in neural networks is sometimes more complex. Small changes to model hyperparameters and initializations can induce the discovery of qualitatively different algorithms from a fixed training set, and even parallel implementations of multiple such algorithms. Some networks trained to perform modular addition implement a familiar Clock algorithm; others implement a previously undescribed, less intuitive, but comprehensible procedure which we term the Pizza algorithm, or a variety of even more complex procedures. Our results show that even simple learning problems can admit a surprising diversity of solutions, motivating the development of new tools for characterizing the behavior of neural networks across their algorithmic phase space.

Ziming Liu, Ouail Kitouni, Niklas Nolte et al.

We aim to understand grokking, a phenomenon where models generalize long after overfitting their training set. We present both a microscopic analysis anchored by an effective theory and a macroscopic analysis of phase diagrams describing learning performance across hyperparameters. We find that generalization originates from structured representations whose training dynamics and dependence on training set size can be predicted by our effective theory in a toy setting. We observe empirically the presence of four learning phases: comprehension, grokking, memorization, and confusion. We find representation learning to occur only in a "Goldilocks zone" (including comprehension and grokking) between memorization and confusion. We find on transformers the grokking phase stays closer to the memorization phase (compared to the comprehension phase), leading to delayed generalization. The Goldilocks phase is reminiscent of "intelligence from starvation" in Darwinian evolution, where resource limitations drive discovery of more efficient solutions. This study not only provides intuitive explanations of the origin of grokking, but also highlights the usefulness of physics-inspired tools, e.g., effective theories and phase diagrams, for understanding deep learning.

Eric J. Michaud, Ziming Liu, Uzay Girit et al.

We propose the Quantization Model of neural scaling laws, explaining both the observed power law dropoff of loss with model and data size, and also the sudden emergence of new capabilities with scale. We derive this model from what we call the Quantization Hypothesis, where network knowledge and skills are "quantized" into discrete chunks ($\textbf{quanta}$). We show that when quanta are learned in order of decreasing use frequency, then a power law in use frequencies explains observed power law scaling of loss. We validate this prediction on toy datasets, then study how scaling curves decompose for large language models. Using language model gradients, we automatically decompose model behavior into a diverse set of skills (quanta). We tentatively find that the frequency at which these quanta are used in the training distribution roughly follows a power law corresponding with the empirical scaling exponent for language models, a prediction of our theory.

Ziming Liu, Eric J. Michaud, Max Tegmark

Grokking, the unusual phenomenon for algorithmic datasets where generalization happens long after overfitting the training data, has remained elusive. We aim to understand grokking by analyzing the loss landscapes of neural networks, identifying the mismatch between training and test losses as the cause for grokking. We refer to this as the "LU mechanism" because training and test losses (against model weight norm) typically resemble "L" and "U", respectively. This simple mechanism can nicely explain many aspects of grokking: data size dependence, weight decay dependence, the emergence of representations, etc. Guided by the intuitive picture, we are able to induce grokking on tasks involving images, language and molecules. In the reverse direction, we are able to eliminate grokking for algorithmic datasets. We attribute the dramatic nature of grokking for algorithmic datasets to representation learning.

Eric J. Michaud, Ziming Liu, Max Tegmark

We explore unique considerations involved in fitting ML models to data with very high precision, as is often required for science applications. We empirically compare various function approximation methods and study how they scale with increasing parameters and data. We find that neural networks can often outperform classical approximation methods on high-dimensional examples, by auto-discovering and exploiting modular structures therein. However, neural networks trained with common optimizers are less powerful for low-dimensional cases, which motivates us to study the unique properties of neural network loss landscapes and the corresponding optimization challenges that arise in the high precision regime. To address the optimization issue in low dimensions, we develop training tricks which enable us to train neural networks to extremely low loss, close to the limits allowed by numerical precision.

Ziming Liu, Varun Madhavan, Max Tegmark

We present a machine learning algorithm that discovers conservation laws from differential equations, both numerically (parametrized as neural networks) and symbolically, ensuring their functional independence (a non-linear generalization of linear independence). Our independence module can be viewed as a nonlinear generalization of singular value decomposition. Our method can readily handle inductive biases for conservation laws. We validate it with examples including the 3-body problem, the KdV equation and nonlinear Schrödinger equation.

Jingcai Guo, Song Guo, Jie Zhang et al.

Federated learning (FL) has emerged as a promising privacy-preserving distributed machine learning framework recently. It aims at collaboratively learning a shared global model by performing distributed training locally on edge devices and aggregating local models into a global one without centralized raw data sharing in the cloud server. However, due to the large local data heterogeneities (Non-I.I.D. data) across edge devices, the FL may easily obtain a global model that can produce more shifted gradients on local datasets, thereby degrading the model performance or even suffering from the non-convergence during training. In this paper, we propose a novel FL training framework, dubbed Fed-FSNet, using a properly designed Fuzzy Synthesizing Network (FSNet) to mitigate the Non-I.I.D. FL at-the-source. Concretely, we maintain an edge-agnostic hidden model in the cloud server to estimate a less-accurate while direction-aware inversion of the global model. The hidden model can then fuzzily synthesize several mimic I.I.D. data samples (sample features) conditioned on only the global model, which can be shared by edge devices to facilitate the FL training towards faster and better convergence. Moreover, since the synthesizing process involves neither access to the parameters/updates of local models nor analyzing individual local model outputs, our framework can still ensure the privacy of FL. Experimental results on several FL benchmarks demonstrate that our method can significantly mitigate the Non-I.I.D. issue and obtain better performance against other representative methods.

Ziming Liu, Pingchuan Ma, Yixuan Wang et al.

A major challenge of AI + Science lies in their inherent incompatibility: today's AI is primarily based on connectionism, while science depends on symbolism. To bridge the two worlds, we propose a framework to seamlessly synergize Kolmogorov-Arnold Networks (KANs) and science. The framework highlights KANs' usage for three aspects of scientific discovery: identifying relevant features, revealing modular structures, and discovering symbolic formulas. The synergy is bidirectional: science to KAN (incorporating scientific knowledge into KANs), and KAN to science (extracting scientific insights from KANs). We highlight major new functionalities in the pykan package: (1) MultKAN: KANs with multiplication nodes. (2) kanpiler: a KAN compiler that compiles symbolic formulas into KANs. (3) tree converter: convert KANs (or any neural networks) to tree graphs. Based on these tools, we demonstrate KANs' capability to discover various types of physical laws, including conserved quantities, Lagrangians, symmetries, and constitutive laws.

Qiyao Liang, Ziming Liu, Mitchell Ostrow et al.

Diffusion models are capable of generating photo-realistic images that combine elements which likely do not appear together in the training set, demonstrating the ability to \textit{compositionally generalize}. Nonetheless, the precise mechanism of compositionality and how it is acquired through training remains elusive. Inspired by cognitive neuroscientific approaches, we consider a highly reduced setting to examine whether and when diffusion models learn semantically meaningful and factorized representations of composable features. We performed extensive controlled experiments on conditional Denoising Diffusion Probabilistic Models (DDPMs) trained to generate various forms of 2D Gaussian bump images. We found that the models learn factorized but not fully continuous manifold representations for encoding continuous features of variation underlying the data. With such representations, models demonstrate superior feature compositionality but limited ability to interpolate over unseen values of a given feature. Our experimental results further demonstrate that diffusion models can attain compositionality with few compositional examples, suggesting a more efficient way to train DDPMs. Finally, we connect manifold formation in diffusion models to percolation theory in physics, offering insight into the sudden onset of factorized representation learning. Our thorough toy experiments thus contribute a deeper understanding of how diffusion models capture compositional structure in data.

Xiaocheng Lu, Ziming Liu, Song Guo et al.

Compositional Zero-Shot Learning (CZSL) aims to recognize novel concepts formed by known states and objects during training. Existing methods either learn the combined state-object representation, challenging the generalization of unseen compositions, or design two classifiers to identify state and object separately from image features, ignoring the intrinsic relationship between them. To jointly eliminate the above issues and construct a more robust CZSL system, we propose a novel framework termed Decomposed Fusion with Soft Prompt (DFSP)1, by involving vision-language models (VLMs) for unseen composition recognition. Specifically, DFSP constructs a vector combination of learnable soft prompts with state and object to establish the joint representation of them. In addition, a cross-modal decomposed fusion module is designed between the language and image branches, which decomposes state and object among language features instead of image features. Notably, being fused with the decomposed features, the image features can be more expressive for learning the relationship with states and objects, respectively, to improve the response of unseen compositions in the pair space, hence narrowing the domain gap between seen and unseen sets. Experimental results on three challenging benchmarks demonstrate that our approach significantly outperforms other state-of-the-art methods by large margins.

Feijie Wu, Song Guo, Zhihao Qu et al.

Compared with full client participation, partial client participation is a more practical scenario in federated learning, but it may amplify some challenges in federated learning, such as data heterogeneity. The lack of inactive clients' updates in partial client participation makes it more likely for the model aggregation to deviate from the aggregation based on full client participation. Training with large batches on individual clients is proposed to address data heterogeneity in general, but their effectiveness under partial client participation is not clear. Motivated by these challenges, we propose to develop a novel federated learning framework, referred to as FedAMD, for partial client participation. The core idea is anchor sampling, which separates partial participants into anchor and miner groups. Each client in the anchor group aims at the local bullseye with the gradient computation using a large batch. Guided by the bullseyes, clients in the miner group steer multiple near-optimal local updates using small batches and update the global model. By integrating the results of the two groups, FedAMD is able to accelerate the training process and improve the model performance. Measured by $ε$-approximation and compared to the state-of-the-art methods, FedAMD achieves the convergence by up to $O(1/ε)$ fewer communication rounds under non-convex objectives. Empirical studies on real-world datasets validate the effectiveness of FedAMD and demonstrate the superiority of the proposed algorithm: Not only does it considerably save computation and communication costs, but also the test accuracy significantly improves.

Ziming Liu, Song Guo, Jingcai Guo et al.

Multi-label zero-shot learning extends conventional single-label zero-shot learning to a more realistic scenario that aims at recognizing multiple unseen labels of classes for each input sample. Existing works usually exploit attention mechanism to generate the correlation among different labels. However, most of them are usually biased on several major classes while neglect most of the minor classes with the same importance in input samples, and may thus result in overly diffused attention maps that cannot sufficiently cover minor classes. We argue that disregarding the connection between major and minor classes, i.e., correspond to the global and local information, respectively, is the cause of the problem. In this paper, we propose a novel framework of unbiased multi-label zero-shot learning, by considering various class-specific regions to calibrate the training process of the classifier. Specifically, Pyramid Feature Attention (PFA) is proposed to build the correlation between global and local information of samples to balance the presence of each class. Meanwhile, for the generated semantic representations of input samples, we propose Semantic Attention (SA) to strengthen the element-wise correlation among these vectors, which can encourage the coordinated representation of them. Extensive experiments on the large-scale multi-label zero-shot benchmarks NUS-WIDE and Open-Image demonstrate that the proposed method surpasses other representative methods by significant margins.

Fushuo Huo, Wenchao Xu, Song Guo et al.

Open-World Compositional Zero-shot Learning (OW-CZSL) aims to recognize novel compositions of state and object primitives in images with no priors on the compositional space, which induces a tremendously large output space containing all possible state-object compositions. Existing works either learn the joint compositional state-object embedding or predict simple primitives with separate classifiers. However, the former heavily relies on external word embedding methods, and the latter ignores the interactions of interdependent primitives, respectively. In this paper, we revisit the primitive prediction approach and propose a novel method, termed Progressive Cross-primitive Compatibility (ProCC), to mimic the human learning process for OW-CZSL tasks. Specifically, the cross-primitive compatibility module explicitly learns to model the interactions of state and object features with the trainable memory units, which efficiently acquires cross-primitive visual attention to reason high-feasibility compositions, without the aid of external knowledge. Moreover, considering the partial-supervision setting (pCZSL) as well as the imbalance issue of multiple task prediction, we design a progressive training paradigm to enable the primitive classifiers to interact to obtain discriminative information in an easy-to-hard manner. Extensive experiments on three widely used benchmark datasets demonstrate that our method outperforms other representative methods on both OW-CZSL and pCZSL settings by large margins.

Ziming Liu, Ziqian Zhong, Max Tegmark

We attribute grokking, the phenomenon where generalization is much delayed after memorization, to compression. To do so, we define linear mapping number (LMN) to measure network complexity, which is a generalized version of linear region number for ReLU networks. LMN can nicely characterize neural network compression before generalization. Although the $L_2$ norm has been a popular choice for characterizing model complexity, we argue in favor of LMN for a number of reasons: (1) LMN can be naturally interpreted as information/computation, while $L_2$ cannot. (2) In the compression phase, LMN has linear relations with test losses, while $L_2$ is correlated with test losses in a complicated nonlinear way. (3) LMN also reveals an intriguing phenomenon of the XOR network switching between two generalization solutions, while $L_2$ does not. Besides explaining grokking, we argue that LMN is a promising candidate as the neural network version of the Kolmogorov complexity since it explicitly considers local or conditioned linear computations aligned with the nature of modern artificial neural networks.

Ziming Liu, Max Tegmark

Neural scaling laws (NSL) refer to the phenomenon where model performance improves with scale. Sharma & Kaplan analyzed NSL using approximation theory and predict that MSE losses decay as $N^{-α}$, $α=4/d$, where $N$ is the number of model parameters, and $d$ is the intrinsic input dimension. Although their theory works well for some cases (e.g., ReLU networks), we surprisingly find that a simple 1D problem $y=x^2$ manifests a different scaling law ($α=1$) from their predictions ($α=4$). We opened the neural networks and found that the new scaling law originates from lottery ticket ensembling: a wider network on average has more "lottery tickets", which are ensembled to reduce the variance of outputs. We support the ensembling mechanism by mechanistically interpreting single neural networks, as well as studying them statistically. We attribute the $N^{-1}$ scaling law to the "central limit theorem" of lottery tickets. Finally, we discuss its potential implications for large language models and statistical physics-type theories of learning.

Ziming Liu, Yixuan Wang, Sachin Vaidya et al.

Inspired by the Kolmogorov-Arnold representation theorem, we propose Kolmogorov-Arnold Networks (KANs) as promising alternatives to Multi-Layer Perceptrons (MLPs). While MLPs have fixed activation functions on nodes ("neurons"), KANs have learnable activation functions on edges ("weights"). KANs have no linear weights at all -- every weight parameter is replaced by a univariate function parametrized as a spline. We show that this seemingly simple change makes KANs outperform MLPs in terms of accuracy and interpretability. For accuracy, much smaller KANs can achieve comparable or better accuracy than much larger MLPs in data fitting and PDE solving. Theoretically and empirically, KANs possess faster neural scaling laws than MLPs. For interpretability, KANs can be intuitively visualized and can easily interact with human users. Through two examples in mathematics and physics, KANs are shown to be useful collaborators helping scientists (re)discover mathematical and physical laws. In summary, KANs are promising alternatives for MLPs, opening opportunities for further improving today's deep learning models which rely heavily on MLPs.

Eric J. Michaud, Isaac Liao, Vedang Lad et al.

We present MIPS, a novel method for program synthesis based on automated mechanistic interpretability of neural networks trained to perform the desired task, auto-distilling the learned algorithm into Python code. We test MIPS on a benchmark of 62 algorithmic tasks that can be learned by an RNN and find it highly complementary to GPT-4: MIPS solves 32 of them, including 13 that are not solved by GPT-4 (which also solves 30). MIPS uses an integer autoencoder to convert the RNN into a finite state machine, then applies Boolean or integer symbolic regression to capture the learned algorithm. As opposed to large language models, this program synthesis technique makes no use of (and is therefore not limited by) human training data such as algorithms and code from GitHub. We discuss opportunities and challenges for scaling up this approach to make machine-learned models more interpretable and trustworthy.

Jinyeop Song, Ziming Liu, Max Tegmark et al.

Neural scaling laws characterize how model performance improves as the model size scales up. Inspired by empirical observations, we introduce a resource model of neural scaling. A task is usually composite hence can be decomposed into many subtasks, which compete for resources (measured by the number of neurons allocated to subtasks). On toy problems, we empirically find that: (1) The loss of a subtask is inversely proportional to its allocated neurons. (2) When multiple subtasks are present in a composite task, the resources acquired by each subtask uniformly grow as models get larger, keeping the ratios of acquired resources constants. We hypothesize these findings to be generally true and build a model to predict neural scaling laws for general composite tasks, which successfully replicates the neural scaling law of Chinchilla models reported in arXiv:2203.15556. We believe that the notion of resource used in this paper will be a useful tool for characterizing and diagnosing neural networks.

David D. Baek, Ziming Liu, Max Tegmark

We present GenEFT: an effective theory framework for shedding light on the statics and dynamics of neural network generalization, and illustrate it with graph learning examples. We first investigate the generalization phase transition as data size increases, comparing experimental results with information-theory-based approximations. We find generalization in a Goldilocks zone where the decoder is neither too weak nor too powerful. We then introduce an effective theory for the dynamics of representation learning, where latent-space representations are modeled as interacting particles (repons), and find that it explains our experimentally observed phase transition between generalization and overfitting as encoder and decoder learning rates are scanned. This highlights the power of physics-inspired effective theories for bridging the gap between theoretical predictions and practice in machine learning.

Jingcai Guo, Qihua Zhou, Ruibing Li et al.

This paper provides a novel parsimonious yet efficient design for zero-shot learning (ZSL), dubbed ParsNets, where we are interested in learning a composition of on-device friendly linear networks, each with orthogonality and low-rankness properties, to achieve equivalent or even better performance against existing deep models. Concretely, we first refactor the core module of ZSL, i.e., visual-semantics mapping function, into several base linear networks that correspond to diverse components of the semantic space, where the complex nonlinearity can be collapsed into simple local linearities. Then, to facilitate the generalization of local linearities, we construct a maximal margin geometry on the learned features by enforcing low-rank constraints on intra-class samples and high-rank constraints on inter-class samples, resulting in orthogonal subspaces for different classes and each subspace lies on a compact manifold. To enhance the model's adaptability and counterbalance over/under-fittings in ZSL, a set of sample-wise indicators is employed to select a sparse subset from these base linear networks to form a composite semantic predictor for each sample. Notably, maximal margin geometry can guarantee the diversity of features, and meanwhile, local linearities guarantee efficiency. Thus, our ParsNets can generalize better to unseen classes and can be deployed flexibly on resource-constrained devices. Theoretical explanations and extensive experiments are conducted to verify the effectiveness of the proposed method.

Qiyao Liang, Ziming Liu, Ila Fiete

Diffusion models are capable of impressive feats of image generation with uncommon juxtapositions such as astronauts riding horses on the moon with properly placed shadows. These outputs indicate the ability to perform compositional generalization, but how do the models do so? We perform controlled experiments on conditional DDPMs learning to generate 2D spherical Gaussian bumps centered at specified $x$- and $y$-positions. Our results show that the emergence of semantically meaningful latent representations is key to achieving high performance. En route to successful performance over learning, the model traverses three distinct phases of latent representations: (phase A) no latent structure, (phase B) a 2D manifold of disordered states, and (phase C) a 2D ordered manifold. Corresponding to each of these phases, we identify qualitatively different generation behaviors: 1) multiple bumps are generated, 2) one bump is generated but at inaccurate $x$ and $y$ locations, 3) a bump is generated at the correct $x$ and y location. Furthermore, we show that even under imbalanced datasets where features ($x$- versus $y$-positions) are represented with skewed frequencies, the learning process for $x$ and $y$ is coupled rather than factorized, demonstrating that simple vanilla-flavored diffusion models cannot learn efficient representations in which localization in $x$ and $y$ are factorized into separate 1D tasks. These findings suggest the need for future work to find inductive biases that will push generative models to discover and exploit factorizable independent structures in their inputs, which will be required to vault these models into more data-efficient regimes.

Isaac Liao, Ziming Liu, Max Tegmark

An essential goal in mechanistic interpretability to decode a network, i.e., to convert a neural network's raw weights to an interpretable algorithm. Given the difficulty of the decoding problem, progress has been made to understand the easier encoding problem, i.e., to convert an interpretable algorithm into network weights. Previous works focus on encoding existing algorithms into networks, which are interpretable by definition. However, focusing on encoding limits the possibility of discovering new algorithms that humans have never stumbled upon, but that are nevertheless interpretable. In this work, we explore the possibility of using hypernetworks to generate interpretable networks whose underlying algorithms are not yet known. The hypernetwork is carefully designed such that it can control network complexity, leading to a diverse family of interpretable algorithms ranked by their complexity. All of them are interpretable in hindsight, although some of them are less intuitive to humans, hence providing new insights regarding how to "think" like a neural network. For the task of computing L1 norms, hypernetworks find three algorithms: (a) the double-sided algorithm, (b) the convexity algorithm, (c) the pudding algorithm, although only the first algorithm was expected by the authors before experiments. We automatically classify these algorithms and analyze how these algorithmic phases develop during training, as well as how they are affected by complexity control. Furthermore, we show that a trained hypernetwork can correctly construct models for input dimensions not seen in training, demonstrating systematic generalization.

Ziming Liu, Yizhou Liu, Eric J. Michaud et al.

We aim to understand physics of skill learning, i.e., how skills are learned in neural networks during training. We start by observing the Domino effect, i.e., skills are learned sequentially, and notably, some skills kick off learning right after others complete learning, similar to the sequential fall of domino cards. To understand the Domino effect and relevant behaviors of skill learning, we take physicists' approach of abstraction and simplification. We propose three models with varying complexities -- the Geometry model, the Resource model, and the Domino model, trading between reality and simplicity. The Domino effect can be reproduced in the Geometry model, whose resource interpretation inspires the Resource model, which can be further simplified to the Domino model. These models present different levels of abstraction and simplification; each is useful to study some aspects of skill learning. The Geometry model provides interesting insights into neural scaling laws and optimizers; the Resource model sheds light on the learning dynamics of compositional tasks; the Domino model reveals the benefits of modularity. These models are not only conceptually interesting -- e.g., we show how Chinchilla scaling laws can emerge from the Geometry model, but also are useful in practice by inspiring algorithmic development -- e.g., we show how simple algorithmic changes, motivated by these toy models, can speed up the training of deep learning models.

Ziming Liu, Eric Gan, Max Tegmark

We introduce Brain-Inspired Modular Training (BIMT), a method for making neural networks more modular and interpretable. Inspired by brains, BIMT embeds neurons in a geometric space and augments the loss function with a cost proportional to the length of each neuron connection. We demonstrate that BIMT discovers useful modular neural networks for many simple tasks, revealing compositional structures in symbolic formulas, interpretable decision boundaries and features for classification, and mathematical structure in algorithmic datasets. The ability to directly see modules with the naked eye can complement current mechanistic interpretability strategies such as probes, interventions or staring at all weights.

Xiaocheng Lu, Ziming Liu, Song Guo et al.

Compositional Zero-shot Learning (CZSL) aims to recognize novel concepts composed of known knowledge without training samples. Standard CZSL either identifies visual primitives or enhances unseen composed entities, and as a result, entanglement between state and object primitives cannot be fully utilized. Admittedly, vision-language models (VLMs) could naturally cope with CZSL through tuning prompts, while uneven entanglement leads prompts to be dragged into local optimum. In this paper, we take a further step to introduce a novel Disentangled and Recurrent Prompt Tuning framework termed DRPT to better tap the potential of VLMs in CZSL. Specifically, the state and object primitives are deemed as learnable tokens of vocabulary embedded in prompts and tuned on seen compositions. Instead of jointly tuning state and object, we devise a disentangled and recurrent tuning strategy to suppress the traction force caused by entanglement and gradually optimize the token parameters, leading to a better prompt space. Notably, we develop a progressive fine-tuning procedure that allows for incremental updates to the prompts, optimizing the object first, then the state, and vice versa. Meanwhile, the optimization of state and object is independent, thus clearer features can be learned to further alleviate the issue of entangling misleading optimization. Moreover, we quantify and analyze the entanglement in CZSL and supplement entanglement rebalancing optimization schemes. DRPT surpasses representative state-of-the-art methods on extensive benchmark datasets, demonstrating superiority in both accuracy and efficiency.

Ziming Liu, Yunyue Chen, Yuanqi Du et al.

Integrating physical inductive biases into machine learning can improve model generalizability. We generalize the successful paradigm of physics-informed learning (PIL) into a more general framework that also includes what we term physics-augmented learning (PAL). PIL and PAL complement each other by handling discriminative and generative properties, respectively. In numerical experiments, we show that PAL performs well on examples where PIL is inapplicable or inefficient.

Ziming Liu, Bohan Wang, Qi Meng et al.

Energy conservation is a basic physics principle, the breakdown of which often implies new physics. This paper presents a method for data-driven "new physics" discovery. Specifically, given a trajectory governed by unknown forces, our Neural New-Physics Detector (NNPhD) aims to detect new physics by decomposing the force field into conservative and non-conservative components, which are represented by a Lagrangian Neural Network (LNN) and a universal approximator network (UAN), respectively, trained to minimize the force recovery error plus a constant $λ$ times the magnitude of the predicted non-conservative force. We show that a phase transition occurs at $λ$=1, universally for arbitrary forces. We demonstrate that NNPhD successfully discovers new physics in toy numerical experiments, rediscovering friction (1493) from a damped double pendulum, Neptune from Uranus' orbit (1846) and gravitational waves (2017) from an inspiraling orbit. We also show how NNPhD coupled with an integrator outperforms previous methods for predicting the future of a damped double pendulum.

Ziming Liu, Guangyu Gao, A. K. Qin et al.

State-of-the-art video action recognition models with complex network architecture have archived significant improvements, but these models heavily depend on large-scale well-labeled datasets. To reduce such dependency, we propose a self-supervised teacher-student architecture, i.e., the Differentiated Teachers Guided self-supervised Network (DTG-Net). In DTG-Net, except for reducing labeled data dependency by self-supervised learning (SSL), pre-trained action related models are used as teacher guidance providing prior knowledge to alleviate the demand for a large number of unlabeled videos in SSL. Specifically, leveraging the years of effort in action-related tasks, e.g., image classification, image-based action recognition, the DTG-Net learns the self-supervised video representation under various teacher guidance, i.e., those well-trained models of action-related tasks. Meanwhile, the DTG-Net is optimized in the way of contrastive self-supervised learning. When two image sequences are randomly sampled from the same video or different videos as the positive or negative pairs, respectively, they are then sent to the teacher and student networks for feature embedding. After that, the contrastive feature consistency is defined between features embedding of each pair, i.e., consistent for positive pair and inconsistent for negative pairs. Meanwhile, to reflect various teacher tasks' different guidance, we also explore different weighted guidance on teacher tasks. Finally, the DTG-Net is evaluated in two ways: (i) the self-supervised DTG-Net to pre-train the supervised action recognition models with only unlabeled videos; (ii) the supervised DTG-Net to be jointly trained with the supervised action networks in an end-to-end way. Its performance is better than most pre-training methods but also has excellent competitiveness compared to supervised action recognition methods.

Ziming Liu, Guangyu Gao, Lin Sun et al.

Small object detection is challenging because small objects do not contain detailed information and may even disappear in the deep network. Usually, feeding high-resolution images into a network can alleviate this issue. However, simply enlarging the resolution will cause more problems, such as that, it aggravates the large variant of object scale and introduces unbearable computation cost. To keep the benefits of high-resolution images without bringing up new problems, we proposed the High-Resolution Detection Network (HRDNet). HRDNet takes multiple resolution inputs using multi-depth backbones. To fully take advantage of multiple features, we proposed Multi-Depth Image Pyramid Network (MD-IPN) and Multi-Scale Feature Pyramid Network (MS-FPN) in HRDNet. MD-IPN maintains multiple position information using multiple depth backbones. Specifically, high-resolution input will be fed into a shallow network to reserve more positional information and reducing the computational cost while low-resolution input will be fed into a deep network to extract more semantics. By extracting various features from high to low resolutions, the MD-IPN is able to improve the performance of small object detection as well as maintaining the performance of middle and large objects. MS-FPN is proposed to align and fuse multi-scale feature groups generated by MD-IPN to reduce the information imbalance between these multi-scale multi-level features. Extensive experiments and ablation studies are conducted on the standard benchmark dataset MS COCO2017, Pascal VOC2007/2012 and a typical small object dataset, VisDrone 2019. Notably, our proposed HRDNet achieves the state-of-the-art on these datasets and it performs better on small objects.

Ziming Liu, Guangyu Gao, Lin Sun et al.

For Convolutional Neural Network-based object detection, there is a typical dilemma: the spatial information is well kept in the shallow layers which unfortunately do not have enough semantic information, while the deep layers have a high semantic concept but lost a lot of spatial information, resulting in serious information imbalance. To acquire enough semantic information for shallow layers, Feature Pyramid Networks (FPN) is used to build a top-down propagated path. In this paper, except for top-down combining of information for shallow layers, we propose a novel network called Image Pyramid Guidance Network (IPG-Net) to make sure both the spatial information and semantic information are abundant for each layer. Our IPG-Net has two main parts: the image pyramid guidance transformation module and the image pyramid guidance fusion module. Our main idea is to introduce the image pyramid guidance into the backbone stream to solve the information imbalance problem, which alleviates the vanishment of the small object features. This IPG transformation module promises even in the deepest stage of the backbone, there is enough spatial information for bounding box regression and classification. Furthermore, we designed an effective fusion module to fuse the features from the image pyramid and features from the backbone stream. We have tried to apply this novel network to both one-stage and two-stage detection models, state of the art results are obtained on the most popular benchmark data sets, i.e. MS COCO and Pascal VOC.