[HN Gopher] Inverted Transformers Are Effective for Time Series ...
       ___________________________________________________________________
        
       Inverted Transformers Are Effective for Time Series Forecasting
        
       Author : beefman
       Score  : 195 points
       Date   : 2023-10-11 18:25 UTC (1 days ago)
        
 (HTM) web link (arxiv.org)
 (TXT) w3m dump (arxiv.org)
        
       | whoa_now wrote:
       | Impressive. What turns ratio are we talking about here?
        
         | smolder wrote:
         | The real question, I think, is autobot or decepticon?
        
       | i-use-nixos-btw wrote:
       | I'm not a ML person so forgive my ignorance here.
       | 
       | It looks interesting, but I'm slightly confused about the way
       | this is presented. It feels like it's coming from the wrong
       | angle.
       | 
       | Specifically, reducing a time series to a sequence of patterns
       | and trying to predict what happens next is something that has
       | been done for decades in some form or another. To me the unique
       | aspect of this is that it fits the approach into a transformer.
       | 
       | So I'd expect to see comparisons against other approaches that do
       | the same thing, not against other transformer approaches.
       | 
       | I wouldn't be confused if the title was "Inverted Transformers
       | are MORE EFFECTIVE THAN NORMAL TRANSFORMERS For Time Series
       | Forecasting".
       | 
       | However, if the target audience are transformer folk then it
       | makes sense, it just seems that I'm looking at it from the other
       | direction.
        
         | sdenton4 wrote:
         | I think we're living in a world where deep learning is winning
         | so consistently that comparison to other methods is often just
         | a time suck. It would be nice to provide a non-DL approach as a
         | baseline, but I would expect it to lag behind the DL methods.
         | 
         | Furthermore, often pre-DL methods can be recast as hand-tuned
         | special cases of DL models - some sequence of linear operations
         | with hand-picked discontinuities sprinkled around. If you can
         | implement the pre-DL method using standard neural network
         | components, then gradient descent training of a neural network
         | "should" find an equivalent or better solution.
        
           | Muller20 wrote:
           | This something that DL researchers like to think but it is
           | definitely not true for time series forecasting. See
           | https://forecastingdata.org/ for some examples where simple
           | non-DL approaches beat state-of-the-art DL systems.
        
           | jldugger wrote:
           | > It would be nice to provide a non-DL approach as a
           | baseline, but I would expect it to lag behind the DL methods.
           | 
           | The M# competitions have usually shown very old forecasting
           | algorithms work quite well, with frankly, way less training
           | overhead and data. Ensemble models usually do best, but for a
           | lot of use cases, DL is probably overkill versus ARIMA or
           | triple exponential smoothing.
        
           | wenc wrote:
           | > I think we're living in a world where deep learning is
           | winning so consistently that comparison to other methods is
           | often just a time suck.
           | 
           | This is quite untrue. DL methods work well when there's a lot
           | of data in closed domains. DL works well by learning from
           | corpuses of text and media where it can make reasonable
           | interpolations.
           | 
           | When you don't have enough data and you don't have a known
           | foundational model that you can do zero shot from, DL doesn't
           | work better than simpler conventional methods.
        
           | Nevermark wrote:
           | _Deep learning models are not better for vast problem areas
           | which have analytical design algorithms. Deep learning 's
           | succession of triumphs has been across areas where analytical
           | design has proven difficult._
           | 
           | First, there are many optimal, or near optimal, direct design
           | algorithms for systems that are well characterized. These
           | solutions are more concise, easier to analyze, reveal
           | important insights, and come with guarantees regarding
           | reliability, accuracy, stability, resource requirements, and
           | operating regimes. Clear advantages over inductively learned
           | solutions.
           | 
           | Second, just _assuming_ that new algorithms are better than
           | older algorithms is completely irrational. An anathema to the
           | purpose and benefits of science, math, and responsible
           | research in general.
           | 
           | If you are going to propose new algorithms, you need to
           | compare the new algorithm against the previous state of the
           | art.
           | 
           | Otherwise practitioners and future researchers will be driven
           | into deadends, deploy pointlessly bad designs, forget
           | important knowledge, and worst of all, lose out on what older
           | algorithms can suggest for improving newer algorithms. With
           | no excuse but gross carelessness.
        
           | wenc wrote:
           | DL also don't win at medium scale tabular data. This paper
           | mentions why and how DL could might better (if it indeed can,
           | with limited sized data)
           | 
           | Why do tree-based models still outperform deep learning on
           | tabular data?
           | 
           | https://arxiv.org/abs/2207.08815
        
         | jandrewrogers wrote:
         | The equivalence relationship between efficient AI and universal
         | sequence prediction has been known for decades, so it would be
         | surprising if AI algorithms were poor at sequence prediction.
         | Of course, _optimal_ universal sequence prediction is
         | profoundly intractable and memory hard, which has implications
         | for limits of AI efficiency and scalability.
         | 
         | There used to be a small hobbyist subculture on the Internet in
         | the late 1990s that designed highly efficient approximate
         | universal sequence predictor algorithms for the challenge of
         | it. Now that AI is a thing, I've often wondered if there were
         | some lost insights there on maximally efficient representations
         | of learning systems on real computers. Most of those people
         | would be deep into retirement by now.
        
           | bravura wrote:
           | There's nothing more fun than dusting off fossilized proto-AI
           | work and running it on modern hardware.
           | 
           | Why don't you share some citations?
           | 
           | I always enjoyed tracking down outre typewritten
           | connectionist manuscripts from an author who had more time
           | than compute.
        
           | Roark66 wrote:
           | Is there anything left of their output in the Internet
           | archive? It is an interesting subject to explore.
        
       | sterlind wrote:
       | Gah, this paper is hard to read, but here's my understanding:
       | 
       | Let's say you have 100 intersections, and you want to predict the
       | traffic on each in cars/sec. You sample every hour, and you keep
       | 24 hours of context, and try to predict the next 4.
       | 
       | First, you'd make 100 "tokens" (really stretching the meaning of
       | token here), one for each stoplight, and loading 24 samples (the
       | history of that stoplight) into each token, and normalize.
       | 
       | Next, you run each token through a Multi-Layer Perceptron
       | (vanilla, old-school neural network) to make a vector of dim D.
       | 
       | Next, for each layer of the transformer, you: 1. Perform "cross-
       | attention," i.e. the query/key/value dance. This is how the
       | different time series (erm, tokens) get to share information. 2.
       | Normalize across all. 3. Run another bog-standard MLP
       | independently on each token. This is the opportunity to examine
       | the history of each time series. 4. Normalize again across all.
       | 
       | Then, you map each "token" (ugh) from being D-dimensional to
       | 4-dimensional, so for each stoplight it predicts the traffic
       | ahead for the next 4 hours. This is also a regular MLP.
       | 
       | So specifically, if you're only predicting a single time series
       | (one stoplight), this method is equivalent to running a regular
       | neural network.
       | 
       | It also, interestingly enough, skips the cool sinusoidal position
       | embedding that transformers use to embed token position. Fair
       | enough, since here the time dimension is fixed and the index of
       | the feed-forward neurons in each MLP layer corresponds (roughly)
       | to the time index of the sample.
       | 
       | The architecture looks weird to me, but apparently it works so
       | that's cool! But I'm not sure how well it works, and my
       | unscientific gut feel is that there's a better and simpler
       | architecture crying out to be found, because this looks a bit
       | tortured. Like, nothing in it explicitly models the time
       | dimension - that task is left to the MLPs - and that seems weird.
        
         | gigatexal wrote:
         | Thank you! Thank you for explaining it in simpler terms. I get
         | about 5% out of these papers but I got a lot more out of this
         | break down.
        
         | fnordpiglet wrote:
         | I find crossformers easier to track:
         | 
         | https://openreview.net/forum?id=vSVLM2j9eie
        
         | benreesman wrote:
         | I had a startup a few years ago that was in the "eh we've got
         | some money left from our BigTech days, let's buy a lottery
         | ticket that's also a masters degree" category.
         | 
         | And in late 2018, attention/transformers was quite the risque
         | idea. We were trying to forecast price action in financial
         | markets, and while it didn't work (I mean really Ben), it
         | smoked all the published stuff like DeepLOB.
         | 
         | It used learned embeddings of raw order books passed through a
         | little conv widget to smooth a bit, and then learned embeddings
         | of order book states before passing them through big-standard
         | positional encoding and multi-head masked self-attention.
         | 
         | This actually worked great!
         | 
         | The thing that kills you is trying to reward-shape on the
         | policy side to avoid getting eaten by taker fees, but it's a
         | broken ATM with artificially lowered fees.
        
           | anonu wrote:
           | why didnt you start a hedge fund?
        
             | dchftcs wrote:
             | They did say it didn't work. The overwhelming majority of
             | finance stuff in published work doesn't work, because
             | they're either too simplistic, poorly backtested, or they
             | get exploited too quickly, so beating those doesn't imply
             | you can run a hedge fund.
             | 
             | The main part here is that it's one thing to predict price
             | action, it's another thing to trade profitably - and in
             | particular they were not able to beat fees, which is a
             | common hurdle if you're new to HFT.
        
               | benreesman wrote:
               | Basically this. We were heavy infra pros and my cofounder
               | was an HFT veteran so it wasn't classic implementation
               | shortfall so much as we didn't solve the "do we enter"
               | threshold on what would be a friction-free windfall.
        
             | Galanwe wrote:
             | What they describe looks like a single predictor. You can't
             | create a strategy with a single predictor, unless it's
             | incredibly predictive. 99% of the time, a predictor cannot
             | beat its transaction costs alone.
             | 
             | You need to combine hundreds of such predictors to be able
             | to beat costs and have a net profitable strategy.
             | 
             | We have a saying in French that you need a lot of rivers to
             | create a sea.
        
               | benreesman wrote:
               | So the group involved veterans from like Knight and DRW
               | and stuff: we understood the model of combining lots of
               | small signals with a low-latency regression.
               | 
               | We were trying to learn those signals as opposed to
               | sweat-shop them.
               | 
               | But the broader point holds: signal isn't alpha.
        
               | nyrikki wrote:
               | Wasn't the US housing crisis of the late 2000s caused by
               | that 99% threshold?
               | 
               | Not in finance at all but I do use reverse Kalman
               | filters, to which this seems similar in core concepts.
               | 
               | While reverse Kalman filters are incredibly helpful in
               | reducing cloud spends by predicting when to auto scale,
               | you still have to have metrics to quickly recover from
               | mistakes.
               | 
               | Based only on tech interviews with HFT companies, I would
               | assume someone could predict your moves using these
               | methods based on historical data.
               | 
               | But perhaps I am just too risk adverse or am missing the
               | core concept.
        
               | benreesman wrote:
               | You might be referring to the Gaussian coppula bullshit
               | Dave Li did? [1]
               | 
               | [1] https://en.m.wikipedia.org/wiki/David_X._Li
        
           | sterlind wrote:
           | Interesting, I'm trying to understand (much less
           | knowledgeable about finance than ML, heh.) But it sounds like
           | you fed it the raw order books (no time dimension), a
           | sequence of order states corresponding to each (a time
           | series), mapped them into the embedding dimension of a
           | decoder-only transformer (the masking), and trained it to
           | predict logits for the next order state?
           | 
           | See, that makes way more sense to me, since it sounds like
           | you used causal self-attention , and actual position
           | embeddings.
           | 
           | I've been interested in some time series stuff, like position
           | embeddings to model actual wall-clock time offsets rather
           | than sequence index, but for textless NLP rather than
           | trading.
        
           | anonymoushn wrote:
           | XTX markets seems to be doing something in the same genre as
           | this. As I understand it, they are mostly taker.
        
         | jameshart wrote:
         | Doesn't this presuppose that all the information you need to
         | predict the future of your time series is embedded in the past
         | of those time series?
         | 
         | Don't _most_ time series we would be interested in predicting
         | (weather, prices, traffic volumes) tend to respond to things
         | outside the history of the time series in question?
         | 
         | Or is the thesis here that we throw every random time series we
         | can think of - wave height series from buoys in the San
         | Francisco Bay, ticket sales from Taylor Swift concerts, Teslas
         | per hour in the Holland tunnel, sales volume of MSFT... and get
         | this thing to find the cross-correlated leading indicators
         | needed so it can predict them all?
        
           | nl wrote:
           | > Doesn't this presuppose that all the information you need
           | to predict the future of your time series is embedded in the
           | past of those time series?
           | 
           | Yes. But usually this is somewhat valid: There might not be
           | data about the causes in your data, but the model _should_
           | learn not be be over confident.
           | 
           | > Don't most time series we would be interested in predicting
           | (weather, prices, traffic volumes) tend to respond to things
           | outside the history of the time series in question?
           | 
           | Yes and no.
           | 
           | You really want the forecast to be a probability
           | distribution: 95% of the time it will take you X minutes to
           | get home from work if you leave at 17:30 but 5% of the time
           | there will be disruptions.
        
             | mirekrusin wrote:
             | Big part of it is historic dice tosses that create mirage
             | of data just waiting to be tamed.
        
         | AndrewKemendo wrote:
         | Really really great write up!
         | 
         | Thank you and yes this is very exciting
         | 
         | People are starting to really decompose the transformer
         | architecture and I'm excited to see how far it can go
        
         | marcyb5st wrote:
         | I think this is a very similar concept compared to TiDE:
         | https://arxiv.org/abs/2304.08424 that also came before and is
         | linked in the paper mentioned in this post. I didn't read
         | through the paper, so I can't point out the differences in
         | approach yet.
         | 
         | However, by just looking at this post' paper results, it seems
         | that at least for TiDE they reported the results completely
         | different from the original paper. It seems this is cherry-
         | picking the particular configuration as the delta is a bit too
         | much to just blame un-reproducibility.
        
       | davesque wrote:
       | Sounds like the basic idea is as follows:
       | 
       | Typical transformers apply self-attention between tokens that
       | vary across time. So the dot product values for each pair of
       | tokens in the resulting attention (correlation) matrix are
       | basically dot products between pairs of moments in time.
       | 
       | The iTransformer authors seem to be saying that, for certain time
       | series forecasting tasks, it's not correct to assume that
       | embedding channels of tokens across moments in time represent
       | data that was collected at precisely the same moment or with
       | similar instruments. In reality, different varieties of data are
       | sometimes not precisely aligned in a data set and also have very
       | different distributions relating to how the data was collected.
       | 
       | So the iTransformer model proposes to apply self-attention across
       | embedding channels instead of across time. Self-attention
       | otherwise seems to work in the same way. Query and key matrices
       | are calculated but they project each embedding channel separately
       | instead of projecting a collection of channel values at a single
       | moment. Then the query-key calculation finds the degree to which
       | all the entirely independent time series (embedding channels) are
       | correlated. Those correlations are used to weight the value
       | vectors and obtain new embedding channels that are weighted
       | averages.
       | 
       | Then the feed-forward layer projects each channel independently,
       | instead of projecting across channels as it would do in a
       | standard transformer model.
       | 
       | Also, since layer normalization acts within an embedding channel,
       | they claim that this can reduce noise that would result from
       | normalizing data across channels that were collected using
       | different methods. The distribution characteristics of each
       | channel stay within the channel instead of bleeding across
       | channels and potentially deleting information.
       | 
       | They lay out more of their reasoning for taking this approach in
       | the paper and I feel like I agree with their intuitions. But the
       | paper needs some serious proof reading. It's very hard to parse
       | the verbiage.
        
       ___________________________________________________________________
       (page generated 2023-10-12 21:02 UTC)