Takayuki Osogami

Nicolò Cesa-Bianchi, Tommaso Cesari, Takayuki Osogami et al.

We study a repeated game between a supplier and a retailer who want to maximize their respective profits without full knowledge of the problem parameters. After characterizing the uniqueness of the Stackelberg equilibrium of the stage game with complete information, we show that even with partial knowledge of the joint distribution of demand and production costs, natural learning dynamics guarantee convergence of the joint strategy profile of supplier and retailer to the Stackelberg equilibrium of the stage game. We also prove finite-time bounds on the supplier's regret and asymptotic bounds on the retailer's regret, where the specific rates depend on the type of knowledge preliminarily available to the players. In the special case when the supplier is not strategic (vertical integration), we prove optimal finite-time regret bounds on the retailer's regret (or, equivalently, the social welfare) when costs and demand are adversarially generated and the demand is censored.

Takayuki Katsuki, Takayuki Osogami

This paper addresses a regression problem in which output label values are the results of sensing the magnitude of a phenomenon. A low value of such labels can mean either that the actual magnitude of the phenomenon was low or that the sensor made an incomplete observation. This leads to a bias toward lower values in labels and the resultant learning because labels may have lower values due to incomplete observations, even if the actual magnitude of the phenomenon was high. Moreover, because an incomplete observation does not provide any tags indicating incompleteness, we cannot eliminate or impute them. To address this issue, we propose a learning algorithm that explicitly models incomplete observations corrupted with an asymmetric noise that always has a negative value. We show that our algorithm is unbiased as if it were learned from uncorrupted data that does not involve incomplete observations. We demonstrate the advantages of our algorithm through numerical experiments.

Takayuki Osogami

This position paper argues that AI agents should be regulated by the extent to which they operate autonomously. AI agents with long-term planning and strategic capabilities can pose significant risks of human extinction and irreversible global catastrophes. While existing regulations often focus on computational scale as a proxy for potential harm, we argue that such measures are insufficient for assessing the risks posed by agents whose capabilities arise primarily from inference-time computation. To support our position, we discuss relevant regulations and recommendations from scientists regarding existential risks, as well as the advantages of using action sequences -- which reflect the degree of an agent's autonomy -- as a more suitable measure of potential impact than existing metrics that rely on observing environmental states.

Ryo Iwaki, Takayuki Osogami

While Distributional Reinforcement Learning (DRL) methods have demonstrated strong performance in online settings, its success in offline scenarios remains limited. We hypothesize that a key limitation of existing offline DRL methods lies in their approach to uniformly underestimate return quantiles. This uniform pessimism can lead to overly conservative value estimates, ultimately hindering generalization and performance. To address this, we introduce a novel concept called quantile distortion, which enables non-uniform pessimism by adjusting the degree of conservatism based on the availability of supporting data. Our approach is grounded in theoretical analysis and empirically validated, demonstrating improved performance over uniform pessimism.

Takayuki Osogami, Hirota Kinoshita, Segev Wasserkrug

We analytically derive a class of optimal solutions to a linear program (LP) for automated mechanism design that satisfies efficiency, incentive compatibility, strong budget balance (SBB), and individual rationality (IR), where SBB and IR are enforced in expectation. These solutions can be expressed using a set of essential variables whose cardinality is exponentially smaller than the total number of variables in the original formulation. However, evaluating a key term in the solutions requires exponentially many optimization steps as the number of players $N$ increases. We address this by translating the evaluation of this term into a multi-armed bandit (MAB) problem and develop a probably approximately correct (PAC) estimator with asymptotically optimal sample complexity. This MAB-based approach reduces the optimization complexity from exponential to $O(N\log N)$. Numerical experiments confirm that our method efficiently computes mechanisms with the target properties, scaling to problems with up to $N=128$ players -- substantially improving over prior work.

Hiroshi Kajino, Kohei Miyaguchi, Takayuki Osogami

We are interested in in silico evaluation methodology for molecular optimization methods. Given a sample of molecules and their properties of our interest, we wish not only to train an agent that can find molecules optimized with respect to the target property but also to evaluate its performance. A common practice is to train a predictor of the target property on the sample and use it for both training and evaluating the agent. We show that this evaluator potentially suffers from two biases; one is due to misspecification of the predictor and the other to reusing the same sample for training and evaluation. We discuss bias reduction methods for each of the biases comprehensively, and empirically investigate their effectiveness.

Takayuki Osogami

We provide complete proofs of the lemmas about the properties of the regularized loss function that is used in the second order techniques for learning time-series with structural breaks in Osogami (2021). In addition, we show experimental results that support the validity of the techniques.

Takayuki Osogami

Here, we provide a supplementary material for Takayuki Osogami, "Uncorrected least-squares temporal difference with lambda-return," which appears in {\it Proceedings of the 34th AAAI Conference on Artificial Intelligence} (AAAI-20).

Kun Zhao, Takayuki Osogami, Tetsuro Morimura

In team-based invasion sports such as soccer and basketball, analytics is important for teams to understand their performance and for audiences to understand matches better. The present work focuses on performing visual analytics to evaluate the value of any kind of event occurring in a sports match with a continuous parameter space. Here, the continuous parameter space involves the time, location, score, and other parameters. Because the spatiotemporal data used in such analytics is a low-level representation and has a very large size, however, traditional analytics may need to discretize the continuous parameter space (e.g., subdivide the playing area) or use a local feature to limit the analysis to specific events (e.g., only shots). These approaches make evaluation impossible for any kind of event with a continuous parameter space. To solve this problem, we consider a whole match as a Markov chain of significant events, so that event values can be estimated with a continuous parameter space by solving the Markov chain with a machine learning model. The significant events are first extracted by considering the time-varying distribution of players to represent the whole match. Then, the extracted events are redefined as different states with the continuous parameter space and built as a Markov chain so that a Markov reward process can be applied. Finally, the Markov reward process is solved by a customized fitted-value iteration algorithm so that the event values with the continuous parameter space can be predicted by a regression model. As a result, the event values can be visually inspected over the whole playing field under arbitrary given conditions. Experimental results with real soccer data show the effectiveness of the proposed system.

Takayuki Osogami, Toshihiro Takahashi

Autonomous agents need to make decisions in a sequential manner, under partially observable environment, and in consideration of how other agents behave. In critical situations, such decisions need to be made in real time for example to avoid collisions and recover to safe conditions. We propose a technique of tree search where a deterministic and pessimistic scenario is used after a specified depth. Because there is no branching with the deterministic scenario, the proposed technique allows us to take into account the events that can occur far ahead in the future. The effectiveness of the proposed technique is demonstrated in Pommerman, a multi-agent environment used in a NeurIPS 2018 competition, where the agents that implement the proposed technique have won the first and third places.

Takayuki Katsuki, Takayuki Osogami, Akira Koseki et al.

This paper proposes a method for modeling event sequences with ambiguous timestamps, a time-discounting convolution. Unlike in ordinary time series, time intervals are not constant, small time-shifts have no significant effect, and inputting timestamps or time durations into a model is not effective. The criteria that we require for the modeling are providing robustness against time-shifts or timestamps uncertainty as well as maintaining the essential capabilities of time-series models, i.e., forgetting meaningless past information and handling infinite sequences. The proposed method handles them with a convolutional mechanism across time with specific parameterizations, which efficiently represents the event dependencies in a time-shift invariant manner while discounting the effect of past events, and a dynamic pooling mechanism, which provides robustness against the uncertainty in timestamps and enhances the time-discounting capability by dynamically changing the pooling window size. In our learning algorithm, the decaying and dynamic pooling mechanisms play critical roles in handling infinite and variable length sequences. Numerical experiments on real-world event sequences with ambiguous timestamps and ordinary time series demonstrated the advantages of our method.

Rudy Raymond, Takayuki Osogami, Sakyasingha Dasgupta

Dynamic Boltzmann Machine (DyBM) has been shown highly efficient to predict time-series data. Gaussian DyBM is a DyBM that assumes the predicted data is generated by a Gaussian distribution whose first-order moment (mean) dynamically changes over time but its second-order moment (variance) is fixed. However, in many financial applications, the assumption is quite limiting in two aspects. First, even when the data follows a Gaussian distribution, its variance may change over time. Such variance is also related to important temporal economic indicators such as the market volatility. Second, financial time-series data often requires learning datasets generated by the generalized Gaussian distribution with an additional shape parameter that is important to approximate heavy-tailed distributions. Addressing those aspects, we show how to extend DyBM that results in significant performance improvement in predicting financial time-series data.

Takayuki Osogami

We review Boltzmann machines and energy-based models. A Boltzmann machine defines a probability distribution over binary-valued patterns. One can learn parameters of a Boltzmann machine via gradient based approaches in a way that log likelihood of data is increased. The gradient and Hessian of a Boltzmann machine admit beautiful mathematical representations, although computing them is in general intractable. This intractability motivates approximate methods, including Gibbs sampler and contrastive divergence, and tractable alternatives, namely energy-based models.

Takayuki Osogami

We review Boltzmann machines extended for time-series. These models often have recurrent structure, and back propagration through time (BPTT) is used to learn their parameters. The per-step computational complexity of BPTT in online learning, however, grows linearly with respect to the length of preceding time-series (i.e., learning rule is not local in time), which limits the applicability of BPTT in online learning. We then review dynamic Boltzmann machines (DyBMs), whose learning rule is local in time. DyBM's learning rule relates to spike-timing dependent plasticity (STDP), which has been postulated and experimentally confirmed for biological neural networks.

Takayuki Osogami

A dynamic Boltzmann machine (DyBM) has been proposed as a model of a spiking neural network, and its learning rule of maximizing the log-likelihood of given time-series has been shown to exhibit key properties of spike-timing dependent plasticity (STDP), which had been postulated and experimentally confirmed in the field of neuroscience as a learning rule that refines the Hebbian rule. Here, we relax some of the constraints in the DyBM in a way that it becomes more suitable for computation and learning. We show that learning the DyBM can be considered as logistic regression for binary-valued time-series. We also show how the DyBM can learn real-valued data in the form of a Gaussian DyBM and discuss its relation to the vector autoregressive (VAR) model. The Gaussian DyBM extends the VAR by using additional explanatory variables, which correspond to the eligibility traces of the DyBM and capture long term dependency of the time-series. Numerical experiments show that the Gaussian DyBM significantly improves the predictive accuracy over VAR.

Sakyasingha Dasgupta, Takayuki Yoshizumi, Takayuki Osogami

We introduce Delay Pruning, a simple yet powerful technique to regularize dynamic Boltzmann machines (DyBM). The recently introduced DyBM provides a particularly structured Boltzmann machine, as a generative model of a multi-dimensional time-series. This Boltzmann machine can have infinitely many layers of units but allows exact inference and learning based on its biologically motivated structure. DyBM uses the idea of conduction delays in the form of fixed length first-in first-out (FIFO) queues, with a neuron connected to another via this FIFO queue, and spikes from a pre-synaptic neuron travel along the queue to the post-synaptic neuron with a constant period of delay. Here, we present Delay Pruning as a mechanism to prune the lengths of the FIFO queues (making them zero) by setting some delay lengths to one with a fixed probability, and finally selecting the best performing model with fixed delays. The uniqueness of structure and a non-sampling based learning rule in DyBM, make the application of previously proposed regularization techniques like Dropout or DropConnect difficult, leading to poor generalization. First, we evaluate the performance of Delay Pruning to let DyBM learn a multidimensional temporal sequence generated by a Markov chain. Finally, we show the effectiveness of delay pruning in learning high dimensional sequences using the moving MNIST dataset, and compare it with Dropout and DropConnect methods.

Takayuki Osogami, Makoto Otsuka

We propose a particularly structured Boltzmann machine, which we refer to as a dynamic Boltzmann machine (DyBM), as a stochastic model of a multi-dimensional time-series. The DyBM can have infinitely many layers of units but allows exact and efficient inference and learning when its parameters have a proposed structure. This proposed structure is motivated by postulates and observations, from biological neural networks, that the synaptic weight is strengthened or weakened, depending on the timing of spikes (i.e., spike-timing dependent plasticity or STDP). We show that the learning rule of updating the parameters of the DyBM in the direction of maximizing the likelihood of given time-series can be interpreted as STDP with long term potentiation and long term depression. The learning rule has a guarantee of convergence and can be performed in a distributed matter (i.e., local in space) with limited memory (i.e., local in time).

Takayuki Osogami

We demonstrate a limitation of discounted expected utility, a standard approach for representing the preference to risk when future cost is discounted. Specifically, we provide an example of the preference of a decision maker that appears to be rational but cannot be represented with any discounted expected utility. A straightforward modification to discounted expected utility leads to inconsistent decision making over time. We will show that an iterated risk measure can represent the preference that cannot be represented by any discounted expected utility and that the decisions based on the iterated risk measure are consistent over time.